Explainin' Video Cards Specs pls.
#1
Posted 06 November 2009 - 08:13 AM
This is question, mind you!
#2
Posted 06 November 2009 - 11:01 AM
coldemone, on 06 November 2009 - 06:13 PM, said:
This is question, mind you!
PCI - Peripheral Component Interconnect, is a 'link' between certain hardware and your motherboard. They have set speeds, the larger the connection the faster the speeds (PCI-E x8 and x16)
MHz - MegaHertz, the value used to determine the speed/transfer rate of a processor whether it be a graphics processor or a normal processor.
GPU - Graphics Processing Unit, aka. The graphics card.
Each graphics card has its own unique architecture, some are more optimized for certain program/games. Like nVidia cards, which have PhysX or CUDA enabled can use the processor to render physics instead of the CPU (It's complex, so this is just a basic meaning).
Pixel Shader - I believe this is the type of technology on a card that enables pixels to be rendered 'better'. Pixel fill rate (How many pixels can be filled per frame/second) is also tied to this.
#3
Posted 06 November 2009 - 11:33 PM
It's good to think of dedicated video cards as mini-computers. They have their own dedicated memory and processor. A couple years ago, their sole purpose was to perform all computations related to rendering any kind of graphics. However, with the introduction of software like Nvidia's CUDA and ATI's Stream, some video cards can also assist your computers main processor with computing. What to look for when shopping:
GPU: Graphics Processing Unit. This is the main processor on a video card. It is specially designed to be efficient at computing graphics-related stuff
Stream Processors: these are cores in the GPU that perform the calculations. The more Stream Processors, the better. (it is important to note that ATI and Nvidia count stream processors differently, so you cannot use this to directly compare them. You can only use this to compare an Nvidia GPU with another nvidia, or ATI with ATI)
ROP/Texture Fill units/pixel shaders: you generally won't need to worry about these, as they go hand in hand with Stream processors (i.e., a video card with a lot of stream processors will usually have more of these)
GPU Speed: measured in Megahertz (MHz) or millions of cycles per second. One cycle is the smallest amount of time the GPU can do anything. The more cycles per second, the faster the GPU.
VRAM: Video Random Access Memory: this is the high-speed memory on-board the card that the GPU uses to store temporary data (such as textures) Generally, dedicated video cards have their own RAM, while integrated video cards use the computer's RAM (which is much slower). Video memory is not the only specification you should consider when buying a video card. A card may have 2 GBs of dedicated memory, but it is useless unless the GPU is powerful enough to utilize all of it (or if you play games that utilize it).
VRAM speed is mainly dependent on its operating speed or frequency (also measured in Megahertz) and type.
Common types of VRAM: GDDR, GDDR2, GDDR3, GDDR5. GDDR stands for Graphics Double Date Rate. The higher the number after the "GDDR", the faster the memory. It is important to note however, high-end Nvidia cards use GDDR3 Memory with high clock speeds. ATI's high-end cards use GDDR5, which has a slower clock speed but sends 4 times as much data per clock. Overall, both series have similar memory speeds.
Memory Bus Width: physically it is the wiring between the memory and gpu and how much data can travel through it at one moment. Measured in bits (there are 8 bits in one byte, so divide the bus width number by 8 and that is how many bytes it can handle). The wider the memory bus, the faster the card will be.
PCI: as Syzygy said, it stands for Peripheral Component Interconnect. Most graphics cards made within the last two years use PCI-express, a faster variant of the original PCI slot. Some manufacturers still make AGP and PCI video cards, but unless you have a computer more than a few years old, you probably have a PCI-express slot. PCI-express 2.0 is currently the fastest, and is standard on most motherboards. It starts out at x1 speed (meaning it has 1 lane), and has x4, x8, or up to x16 speed. Most video cards use PCI-Express 2.0 x16.
But honestly, knowing these terms is only part of it. If you plan on looking at graphics cards, I suggest going to Toms Hardware and looking at the benchmarks of various cards.
Here are some video cards by ATI and Nvidia, based on usage. In general, cost is based on performance. You get what you pay for.
Low-end (average person; adequate for basic computer usage and older games): ATI Radeon HD 3000 series, HD 4550. Nvidia Geforce GT 220, 9400 or 9500
Mid-range (play games like CoD4 or Company of Heroes on high settings, but not the newest games): ATI 4650/4670, Nvidia Geforce 9800GT or 9800GTX
High-end: (play most new games on high settings at higher resolutions): ATI Radeon 4850/4870/4890, Radeon 5770. Nvidia GTX 260 (core 216) and GTX 275
Enthusiast: (play most new games on max settings at high resolutions) ATI Radeon 5850/5870, 4870X2. Nvidia GTX 285/295
Currently, the ATI Radeon 5870 is the most powerful single-GPU video card. Nvidia's Geforce GTX 295 is the fastest single card, but it has two GPUs. In some memory-intensive games, the 5870 is faster due to it's wider memory bus.
Hope this helps.
#4
Posted 07 November 2009 - 12:01 AM
Since Guvna Guy and Syzygy hit the nail on top of the head...
RAM - the best way to understand it is simple:
MHz - MegaHertz - speed
MB - Megabyte - amount
If you ever wanted to upgrade, check out: Crucial's website
You can download The Crucial System Scanner Tool and The Crucial Memory Advisor Tool to help you with your upgrade.
Both of these tools will tell you exactly what RAM is currently on your system (Mhz and MB) and tell you what you can upgrade to.
I would always try to Max out the MHz and then stick to at least 2048 MB as a minimum for whatever OS you are running.
Confirm your findings with your OEM and then shop it out - RAM is dirt cheap these days...
If you are upgrading the Video Card, I'd also think about your Processor as well.
poolsharkzz
Need_a_Wash.gif (30.42K)
Number of downloads: 0
#6
Posted 07 November 2009 - 02:10 PM
Can't understand though why my dad says a 800 Mhz in a vcard cannot be used in a computer with less... i dunno, 400 MHz of a something in the computer so he can't use the card in our other old computer to put in the new computer.
Edit: wait, RAM has a MHz rate?
This post has been edited by coldemone: 07 November 2009 - 02:12 PM
#7
Posted 07 November 2009 - 04:38 PM
coldemone, on 07 November 2009 - 10:10 AM, said:
Can't understand though why my dad says a 800 Mhz in a vcard cannot be used in a computer with less... i dunno, 400 MHz of a something in the computer so he can't use the card in our other old computer to put in the new computer.
Edit: wait, RAM has a MHz rate?
That's odd. Are you sure the "400" isn't referring to the capacity of the computer's power supply? Most mid-range cards now require at least a 450 watt power supply for the computer. Other than that the only compatibility issue I would see is if the slot of the new card doesn't match any on the old motherboard.
And yes, all types of RAM have a MHz rate. the faster this rate, the faster your memory can transfer data to and from the processor. The speed of system RAM and VRAM are generally advertised differently though. With VRAM, they usually give you the actual clock speed of the VRAM, whereas your computer's RAM is advertised as DDR2 800, or DDR3 1333 or something like that. When you see, for example, DDR2 800 RAM, the "800" is the amount of data transfers per second (in this case, 800 million). I believe the base clock for DDR2 800 is 200MHz, with an I/O speed output of 400MHz. Sometimes you'll see it advertised as DDR2 800MHz memory but that's a misnomer. However, you don't have to worry about those numbers, as all DDR2 800 Memory sticks have the same frequency. Just like all DDR2 1066 memory sticks have the same frequency as other DDR2 1066. Same goes for DDR3 1333, DDR3 1600, and so on.
#8
Posted 07 November 2009 - 06:16 PM
Most of them should be accurate.
#9
Posted 07 November 2009 - 06:52 PM
coldemone, on 07 November 2009 - 09:10 PM, said:
Can't understand though why my dad says a 800 Mhz in a vcard cannot be used in a computer with less... i dunno, 400 MHz of a something in the computer so he can't use the card in our other old computer to put in the new computer.
Edit: wait, RAM has a MHz rate?
I think your dad is referring to your PSU (Power Supply Unit). Unless you are overclocking, like Guvna Guy said, 450 watt is enough for mid-range GPU's.
#10
Posted 07 November 2009 - 07:15 PM
#11
Posted 07 November 2009 - 07:20 PM
hab, on 08 November 2009 - 02:15 AM, said:
I think we are talking about GPU's here!?
#12
Posted 07 November 2009 - 07:32 PM
hab, on 07 November 2009 - 03:15 PM, said:
Like GPU's you have to look at benchmarks to really know which ones are better. Clock speed is important, but it's not the only factor. Number of cores, as well as the amount of cache the processor has will also make a difference. For example, in multi-threaded apps (like most new games), a 2GHz quad-core will generally be faster than a 2.5GHz dual-core. However, in single threaded apps clock speed is key. Also, the micro-architecture of Intel's Core processors (in Core 2, i5, and i7) is faster clock-per-clock than AMD's, so you can't compare them solely on clock speed.
Also, Tubro Boost, which is a feature on Intel's Core i7 and i5 processors, will give you better performance in single-threaded applications (think of it as "past-proofing" the processor). Of course, AMD's processors offer a very good price/performance ratio. They are perfectly fine for the casual gamer.
Just go to Tom's Hardware, that site has a lot of benchmark data for CPUs and GPUs. They even do all the hard work and tell you what the best CPUs and GPUs for your money are.
http://www.tomshardware.co.uk/
#13
Posted 07 November 2009 - 07:39 PM
fediaFedia, on 07 November 2009 - 02:16 PM, said:
Most of them should be accurate.
Are you sure those are accurate? It has the Nvidia Geforce GTX 295 listed as 11th place but it's still the fastest single graphics card in most cases. Although at high resolutions the ATI Radeon HD 5870 is faster because it has a larger 1GB frame buffer as opposed to the 295's 896MB buffer (technically it's 896MB X 2 for 1.792GB but because this is split between 2 GPUs it makes a bottleneck of 896)
#14
Posted 07 November 2009 - 08:21 PM
Guvna Guy, on 07 November 2009 - 09:39 PM, said:
fediaFedia, on 07 November 2009 - 02:16 PM, said:
Most of them should be accurate.
Are you sure those are accurate? It has the Nvidia Geforce GTX 295 listed as 11th place but it's still the fastest single graphics card in most cases. Although at high resolutions the ATI Radeon HD 5870 is faster because it has a larger 1GB frame buffer as opposed to the 295's 896MB buffer (technically it's 896MB X 2 for 1.792GB but because this is split between 2 GPUs it makes a bottleneck of 896)
People who use Passmark Performance test get accurate results, and then you can submit it to their database, many people submit their results, so the final score is calculated from many. Configurations are different, so performance can be different too. So in this case, the average score for GTX 295 gets 11th place.
#15
Posted 07 November 2009 - 08:23 PM
For the GPU- I'm basically looking for something like a dual core 3.0 Ghz
This post has been edited by hab: 07 November 2009 - 08:28 PM

Sign In
Register
Help
Add Reply


MultiQuote