Graphics cards - The gamer's delight! - Updated 4-04-02
Apr 13 '01 (Updated Apr 04 '02)
The Bottom Line What is the difference between graphics cards??
There's no denying the importance of the graphics card in a computer gaming system, but with all the choices out there what's the difference? Price is a main concern and paying more usually gets you more, but only from the same vendor. And even then you have to be careful! The shopper armed with knowledge is going to get the best deal, but shopping can be made easier by knowing what you want and need. Today we all need more than adequate 2D and 3D performance, but not everyone can afford a $500 graphics card even though we want one :)
In this epinion, I'll focus on the basics of modern graphics cards as well as give you the 'tech specs', which should be useful to hardware afficionados (who will probably be the only ones to appreciate my collection of specs). For more general advice, read the other graphics cards epinions, and I may write one myself.
====================================================
Update: Fixed some ATI errors - it's difficult to get reliable information on their graphics cards because they seem reluctant to release it! Their website has almost zero info. Also deleted most of the older graphics cards and added some new ones to both ATI and NVidia, including the hot GeForce4! Hopefully these new ATI specs are correct, or at least closer to correct ;)
====================================================
What do I need to know? - Skip this if you already know it!
RAM
Current graphics cards have 32MB to 128MB of RAM on them, either SDR (single data rate) or DDR (double data rate). Any serious gamer these days should be able to afford a 64MB card, if not one of the awesome 128MB cards. DDR RAM is RAM that moves twice as much data in a single clock cycle as SDR RAM or 'normal' RAM, meaning that 100MHz DDR RAM is just as fast as 200MHz SDR RAM (ignoring latencies and such). This has taken graphics cards to new heights because an increase in the RAM clock speed results in a double performance increase for DDR RAM! Fast-paced 3D games (as if there's any other kind of game ;) show the biggest performance gain with DDR RAM because of the speed required to render all the frames of animation. I suggest a DDR-based graphics card because game performance is much better and price isn't much more than SDR. I'll leave the SDR-based graphics cards to those who need a basic computer system under $800.
GPU
The 'graphics processor unit', coined by NVidia after the CPU or main computer processor, is a good term for the number-crunching chips on current graphics cards. Past graphics cards used a basic collection of transistors for hardware acceleration of 2D and 3D images, but current graphics cards use full-blown processors that are about as powerful as a Pentium4!
Chipsets and Companies
ATI
The only real 3D gaming competitor to NVidia, ATI has a few tricks up its sleeve, including reportedly better 2D performance, which is especially important when running Windows and, well, most applications when you think about it! ATI makes their own cards, but they also allow third party manufacturers to produce 'OEM' or 'Powered by ATI' cards so watch out for those! ATI does offer a wide range of graphics cards from the basic gamer's set-up to dual-monitor and TV-in/out solutions. Their 'All-in-Wonder' cards are especially useful as a low-end graphics card and a video center for porting TV and analog video into (and out of) your computer for easy editing. Great fun, but these are also rather expensive. I should note, however, that ATI's driver support has historically been less than stellar and this round is no different, in Windows 2000 and XP mostly, though this is constantly improving.
NVidia
NVidia has long been a gamer's best friend, and for good reason! Their product support is superb and all of their drivers are standardized (which means rock-solid performance in every Windows environment). This is most likely because they don't release any graphics cards themselves, but sell the chipset to other manufacturers. These manufacturers then decide what to put on the card, whether it be full digital video editing support or a basic gaming card or (my personal fav) bundled games :) So on the one hand you have more universal support for your NVidia-based graphics card(s), but on the other hand you have to shop around for the right card from the right manufacturer that will fit your needs and budget.
STMicro
STMicro is not a well-known company, but their graphics technology, PowerVR, is. Recognize it? Yes, it's the same technology behind the respectable Dreamcast gaming console, and it has also been available in value graphics cards since 1999. However, value options from NVidia and ATI have typically been better, slightly more expensive, and much more advertised. The third iteration of PowerVR technology has hit the market with a storm thanks to graphics card manufacturer Hercules (3D Prophet 4500). The Kyro II GPU is a fairly powerful and cheap alternative to the likes of NVidia and ATI.
====================================================
Pixels and Texels: Fillrate
I think everyone knows what a pixel is, but a texel? A texel is a textured pixel meaning that detail is added to it, instead of a bland polygon-like look. Trust me, if you play games, you'll appreciate textures if you haven't already seen them. They really add a lot of realism to a scene and are more easily recognised by the human eye. With that said, you need a lot of textures in your games! Thanks to recent advancements in GPU's, textures are actually more plentiful than pixels so you won't run out of them :) The number of pixels and texels that can be generated by the GPU in one second is termed the 'fillrate' of that GPU (related to how well it can 'fill' your monitor with images). Fillrate is now measured in hundreds of millions and even billions of pixels and texels per second, but don't expect this much all the time or even half the time. Since this spec is merely a measure of maximum potential, if everything works to its fullest, it's really only useful to compare fillrates rather than to choose a graphics card based solely on fillrate.
Polygons, Triangles, and Anti-Aliased Samples?
All polygons are made up of one or more triangles, but polygon sounds better so NVidia uses that word when they really mean the simplest polygon - a triangle - just like ATI measures. Actually, because of the wide disparity in triangles per second specs, I suspect that ATI is really measuring a bit more than NVidia. As for 'anti-aliased samples', as far as I have been able to calculate, there are about 28.8 million triangles for every one billion 'anti-aliased samples'.
The Contendors - Who's a player?
ATI's GPU designs are the Radeon 7000 Series (30 million transistors), and the all-powerful Radeon 8500 Series (60 million transistors). NVidia's are the GeForce2 Series (25 million transistors), the GeForce3 Series (57 million transistors), the GeForce4MX Series, and the GeForce4 Ti Series (63 million transistors). STMicro's Kyro II is a small GPU (only 15 million transistors), but its performance is surprising thanks to its tile-based reduced rendering technology.
====================================================
From ATI
Radeon 7000 Series: 256-bit GPU, 0.18 or 0.15 micron process, 30 million transistors, 128-bit or 64-bit memory bus width, hardware DVD decoding (with ATI's special DVD player), 350MHz RAMDAC, and 3D imaging capabilities up to 2048 x 1536.
Radeon 7000 - 0.18 micron process
183MHz GPU core speed, 32MB or 64 MB of 183MHz RAM (128-bit SDR or 64-bit DDR)
183 Megapixels per second, 550 Megatexels per second
14 million triangles per second
2.9 GB per second memory bandwidth
One pipeline with 3 textures per pipeline
- Recommendation: This is clearly the lowest-end graphics card that ATI offers and it shows! With less power than the feeble GeForce2 MX 200, this card is only for those who want better graphics performance than onboard video for their older games. Or maybe for a second computer...
Radeon 7200 - 0.18 micron process
166MHz GPU core speed, 32MB or 64MB of 166MHz RAM (128-bit SDR)
333 Megapixels per second, 1 Gigatexel per second
25 million triangles per second
2.7 GB per second memory bandwidth
Two pipelines with 3 textures per pipeline
- Recommendation: Slower than the 7000 in pure speed, but having the full two pipelines sure does boost performance! And that performance isn't bad, just not up to snuff for the latest games. Again, this would be great for a second computer.
Radeon 7500 (RV200) - All-in-Wonder, 0.15 micron process
260MHz GPU core speed, 64MB of 180MHz DDR RAM (360MHz effectively)
520 Megapixels per second, 1.56 Gigatexels per second
40 million triangles per second
5.8 GB per second memory bandwidth
Two pipelines with 3 textures per pipeline
Dual 400MHz RAMDAC
S-Video, Analog/Component video ports, in and out
- Recommendation: No Firewire ports like the Radeon 8500DV A-i-W, but analog ports are sufficient for such a low price and they both share two of the fastest RAMDAC's I've seen on a consumer-level graphics card, so keeping that MPEG-2 video file at full frame rates should be no problem. Memory bandwidth suffers yet again, but compared to the 7000 and 7200 this All-in-Wonder has what it takes to play games rather respectably!
Radeon 7500 (RV200) - 0.15 micron process
290MHz GPU core speed, 64MB of 230MHz DDR RAM (460MHz effectively)
580 Megapixels per second, 1.74 Gigatexels per second
45 million triangles per second
7.4 GB per second memory bandwidth
Two pipelines with 3 textures per pipeline
- Recommendation: The main hitter for ATI in the mid-range ballpark, this graphics card trades blows with the GeForce2 Ti, often coming out on top. The one thing it does come out underneath, however, is price! Beat that, NVidia!
Radeon 8500 Series (R200): 256-bit GPU, 0.15 micron process, 60 million trasistors, 128-bit memory bus width, hardware DVD decoding (with ATI's special DVD player), 350MHz RAMDAC, 3D imaging capabilities up to 2048 x 1536, and HRAA capabilities.
Radeon 8500DV - All-in-Wonder
230MHz GPU core speed, 64MB of 190MHz DDR RAM (380MHz effectively)
1.08 Gigapixels per second, 2.17 Gigatexels per second
62.8 million triangles per second
6.1 GB per second memory bandwidth
Four programmable pipelines with 2 textures per pipeline
Dual 400MHz RAMDAC
S-Video, Firewire/IEEE 1394, and Component video ports, in and out
- Recommendation: The memory bandwidth on this 8500 suffers severely compared to standard specs. However, this graphics card is an 'All-in-Wonder' so video input/output and manipulation are its thing; again, take a look at the speedy dual RAMDAC's. And it does it well, especially with a much needed Firewire port for digital cameras and camcorders!
Radeon 8500LE
230MHz GPU core speed, 64MB of 230MHz DDR RAM (460MHz effectively)
1.08 Gigapixels per second, 2.17 Gigatexels per second
62.8 million triangles per second
7.4 GB per second memory bandwidth
Four programmable pipelines with 2 textures per pipeline
- Recommendation: Almost a contradiction in existence, the 8500LE is a value graphics card based on a top-of-the-line GPU! For those who can't afford a full-powered 8500 but want their vertexes shaded a la DirectX 8.1, the LE version may be for you.
Radeon 8500 (original)
275MHz GPU core speed, 64MB of 275MHz DDR RAM (550MHz effectively)
1.3 Gigapixels per second, 2.6 Gigatexels per second
75 million triangles per second
8.8 GB per second memory bandwidth
Four programmable pipelines with 2 textures per pipeline
- Recommendation: Now known as the 'original' 8500, the 64MB version is certainly powerful enough to handle any game on the market. And of course, that nifty DirectX 8.1 support makes everything look all pretty :)
Radeon 8500 128
275MHz GPU core speed, 128MB of 275MHz DDR RAM (550MHz effectively)
1.3 Gigapixels per second, 2.6 Gigatexels per second
75 million triangles per second
8.8 GB per second memory bandwidth
Four programmable pipelines with 2 textures per pipeline
- Recommendation: With twice the video memory of the 'original' Radeon 8500, which should prove useful for this summer's programmable shader games, this is the current top dog in ATI's pack. Current games show almost no performance gain with 128MB of video memory, but then again current games have little to no programmable shader support, either. The grapevine is full of rumors about DirectX 8/8.1 games being released this summer - coincidence?
====================================================
From NVidia
GeForce2 Series: 256-bit GPU, 0.18 micron process, 25 million transistors, 64-bit or 128-bit memory bus width, 350MHz RAMDAC, and 3D imaging capabilities up to 2048 x 1536.
GeForce2 MX 400 (NV 11)
200MHz GPU core speed, 32MB or 64MB of 166MHz 6ns DDR RAM (64-bit bus width, 333MHz effectively)
400 Megapixels per second, 800 Megatexels per second
25 million triangles per second
2.7 GB per second memory bandwidth
Two pipelines with 2 textures per pipeline
- Recommendation: The MX series has a replacement already? Tell that to Compaq and Sony, who still sell the antiquated TNT2 M64 with their "multimedia PCs"! The MX 400 is still the best of the GF2 MX series, but the only one I'm still recommending to anyone who plays modern 3D games. One look at the specs and you'll know why, but mainly it's because you can pick up a retail version for about the cost of the latest game; ONE GAME, quite a steal for any graphics card.
GeForce2 Ti (NV15) (Titanium)
250MHz GPU core speed, 64MB of 200MHz 5ns DDR RAM (400MHz effectively)
1 Gigapixel per second, 2 Gigatexels per second
31 million triangles per second
6.4 GB per second memory bandwidth
Four pipelines with 2 textures per pipeline
- Recommendation: NVidia has so many darn GPU's right now that it's difficult to pick out a place to put this one. How about the lower end of mid-range? If you don't want the humiliation of having a low-end card but are still strapped for cash (and willing to hold out on DirectX 8/8.1 until the majority of games use it), then this would be the card for you. More than acceptable performance for all games, as long as you can do without anti-aliasing.
GeForce4 MX Series (NV17): 256-bit GPU, 0.15 micron process, 29 million transistors, 128-bit memory bus width, 350MHz RAMDAC, 3D imaging capabilities up to 2048 x 1536, and HRAA capabilities. Warning! Although the GeForce4 MX series has a '4' in its name, it is not related to the GeForce4 Ti GPU's at all, and is, in fact, merely a replacement for the GeForce2 MX series of GPU's. No programmable pixels shaders or DirectX 8 support.
GeForce4 MX 420
250MHz GPU core speed, 64MB of 166MHz SDR RAM
500 Megapixels per second, 1 Gigatexel per second
31 million triangles per second
2.7 GB per second memory bandwidth
Two pipelines with 2 textures per pipeline
- Recommendation: As a replacement for the GF2 MX series, this one rocks the house! Hindered by memory bandwidth, yes, but nonetheless impressive for its price. You could probably even play the latest games... at 1024 x 768 or lower, however.
GeForce4 MX 440
270MHz GPU core speed, 64MB of 200MHz DDR RAM (400MHz effectively)
550 Megapixels per second, 1.1 Gigatexels per second
34 million triangles per second
6.4 GB per second memory bandwidth
Two pipelines with 2 textures per pipeline
- Recommendation: A value card with full 128-bit DDR memory? Hello, Dolly! Gaming on the cheap has never been this good! For a little more green than the MX 400 you can now get about 4 times the performance. This was pretty much inconceivable a year ago, when the GeForce2 Ultra was still the king. Geez, has it only been a year?...
GeForce4 MX 460
300MHz GPU core speed, 64MB of 275MHz DDR RAM (550MHz effectively)
600 Megapixels per second, 1.2 Gigatexels per second
38 million triangles per second
8.8 GB per second memory bandwidth
Two pipelines with 2 textures per pipeline
- Recommendation: Better than the MX 440 (and possibly the GF2 Ti), to be sure, which says a lot, but the current pricing scale leaves much to be desired! At roughly $30 less than a GF3 Ti200, it's difficult to justify the cost of this value, non-vertex shading, lack of performance kind of graphics card. Wait for prices to go down or just spring for the extra $30 and get a GF3 Ti200 and enjoy the glow of 3 times the performance!
GeForce3 Series (NV20): 256-bit GPU, 0.15 micron process, 60 million transistors, 128-bit memory bus width, 350MHz RAMDAC, 3D imaging capabilities up to 2048 x 1536, and HRAA capabilities.
GeForce3 Ti 200
175MHz GPU core speed, 64MB of 200MHz 5ns DDR RAM (400MHz effectively)
700 Megapixels per second, 1.4 Gigatexels per second
80 million triangles per second, 2.8 billion anti-aliased samples per second
6.4 GB per second memory bandwidth
Four programmable pipelines with 2 textures per pipeline
2X and 4X FSAA up to 1024 x 768
- Recommendation: An itty-bitty value version of the standard GF3. Despite its specs, the GF3 is designed to be more powerful than a GF2 (or GF4 MX) of similar speed and can make better use of memory bandwidth. Excellent performance for intense gaming, excellent mid-range price :)
GeForce3
200MHz GPU core speed, 64MB of 230MHz 4ns DDR RAM (460MHz effectively)
800 Megapixels per second, 1.6 Gigatexels per second
92 million triangles per second, 3.2 billion anti-aliased samples per second
7.4 GB per second memory bandwidth
Four programmable pipelines with 2 textures per pipeline
2X and 4X FSAA (see below) up to 1024 x 768
- Recommendation: The hottest buzzwords in the graphics market today are "programmable vertex shading" and it all started with this baby! Anyone who owns the original GF3 has never regretted the purchase; even after 9 months (almost a full generation in the insane graphics technology market), this is still a major benchmarking card. Of course, it's naturally cheaper now than it was 9 months ago.
GeForce3 Ti 500
240MHz GPU core speed, 64MB of 250MHz 4ns DDR RAM (500MHz effectively)
960 Megapixels per second, 1.9 Gigatexels per second
109 million triangles per second, 3.8 billion anti-aliased samples per second
8 GB per second memory bandwidth
Four programmable pipelines with 2 textures per pipeline
2X and 4X FSAA up to 1024 x 768
- Recommendation: These 'Ultra' versions are usually the first to go! By that I mean the new 128MB graphics cards (such as the GF4 Ti's) have pretty much taken over the market formerly held by the GF3 Ti 500. Not that this is a bad card anymore, just that its price/performance ratio has been shot to swiss cheese! If you see one of these cards, make sure it's the same price as a GF4 Ti 4200, otherwise I recommend you skip it. For those who already have one, I'm jealous.
GeForce4 Ti Series (NV25): 256-bit GPU, 0.15 micron process, 63 million transistors, 128-bit memory bus width, dual 350MHz RAMDAC, 3D imaging capabilities up to 2048 x 1536, and HRAA capabilities. Notice the dual RAMDAC? The GF4 Ti GPU has built-in support for an extra monitor or LCD screen by default! The GF4 Ti GPU also includes another programmable vertex shader, bringing the total to two (the GF3 had one).
GeForce4 Ti 4200
225MHz GPU core speed, 128MB of 250MHz DDR RAM (500MHz effectively)
900 Megapixels per second, 1.8 Gigatexels per second
102 million triangles per second, 3.6 billion AA samples per second
8 GB per second memory bandwidth
Two programmable pipelines with 2 textures per pipeline
- Recommendation: Well, well, if NVidia didn't just set another generation speed record! And here I thought that it was supposed to take a year to come out with another GPU! 9 months, you say? That's so insane that by the time the GF3 Ti's were available in quantities, NVidia hits us with the aptly named GeForce4 Ti! And at an MSRP of $200, it goes without saying that if you can afford one, get one!
GeForce4 Ti 4400
275MHz GPU core speed, 128MB of 275MHz DDR RAM (550MHz effectively)
1.1 Gigapixels per second, 2.2 Gigatexels per second
125 million triangles per second, 4.4 billion AA samples per second
8.8 GB per second memory bandwidth
Two programmable pipelines with 2 textures per pipeline
- Recommendation: Next in line we have the mandatorily faster, higher model numbered bigger brother of the 4200, the 4400! Uh oh, I think we're starting to leave the Radeon 8500 128 behind! Somebody get a wheelchair and hook it up to the rear bumper of the 4400! Let's hope it doesn't get wind-whipped. MSRP: $300, baby!
GeForce4 Ti 4600
300MHz GPU core speed, 128MB of 325MHz DDR RAM (650MHz effectively)
1.2 Megapixels per second, 2.4 Gigatexels per second
136 million triangles per second, 4.8 billion AA samples per second
10.4 GB per second memory bandwidth
Two programmable pipelines with 2 textures per pipeline
- Recommendation: Cost of admission to the most sensual vertex shading display today: $400. The price is starting to hurt, and don't forget that the GF3 Ti 500 was put to pasture by this version's little brother in less than 6 months. Still, for those who forked over major moolah for the GF2 Ultra and are looking for another 'Ultra' replacement, this one's for you!
====================================================
From STMicro
Kyro II GPU: 256-bit GPU, 0.18 micron process, 15 million transistors, 128-bit memory bus width, and 270MHz RAMDAC.
Kyro II
175MHz GPU core speed, 32MB or 64MB of 175MHz 5.5ns or 5ns SDR RAM
350 Megapixels per second, 700 Megatexels per second
2.8 GB per second memory bandwidth
Two pipelines with 2 textures per pipeline
Limited to 2X AGP mode
2X and 4X FSAA up to 1024 x 768
- Recommendation: Grew up in the 1970's? Then fight the man with the only graphics card to NOT feature any ATI or NVidia influence. Also recommended for nostalgic owners of the Sega Dreamcast. For serious gamers on a budget, the Kyro II is a worthy contestant for your consumership, but you also won't be playing Unreal Tourney 2 on this card if you can help it.
====================================================
As you can see, NVidia usually has more graphics chipset options than its competitors, whereas ATI has always relied on simplicity to survive. Since these two companies are all that's left of a once varied graphics hardware market, both strategies must work. Then there's STMicro with their Kyro II as the black (but somewhat capable) sheep of graphics cards. While it's certainly being left far behind by the new GeForce4 Ti and Radeon 8500, STMicro may already have a Kyro III in the works. NVidia and ATI better watch their backs because even a sufficiently fast Kyro II could compete in the mid-range arena. For now, however, the Kyro line is geared towards the value end of the graphics market, which at the same time means increased competition and thus more power for those of you with less than $100 to drop on a single card. If STMicro does continue their presence in the graphics card market, things could get interesting over the next 18 months. We could very well see better than GeForce3 performance for less than $100 by the dawn of 2003 :)
Final Word
The bottom line is that if you want to say "programmable vertex shading" to your friends and watch them turn green, this is the time to buy into the DirectX 8/8.1 hype. Prices never looked so good! If you've got a 16MB or less TNT2 or Voodoo3 and want more fps out of QuakeIII, now is the time to upgrade; you'll see significant performance increases for many of your 3D games, if not all. If you've got a 2MB Trident PCI graphics card (or less!), steal daddy's credit card and buy yourself a DDR board now!
====================================================
Glossary of little-used terms
This is where I will put the little details on graphics cards and their technologies that don't easily fit into my 'basic guide'. If anyone has suggestions for this section, leave a comment.
FSAA
Full-scene anti-aliasing is something that the former 3dfx graphics company pioneered on PCs, and it looks great! Basically, it smooths angled lines, eliminating the 'stair-case look'. Pixels are tiny rectangles that are hardly noticeable, until you start playing 3D games and look at a straight line at an angle. An angled line is made up of shorter lines that are moved over one pixel from eachother, creating a boxy look that has been coined the 'stair-case look' because of the jagged edges. Anti-aliasing smooths the jagged edges at the affected pixels by creating a softer transition from one pixel line to the next. The result is amazing, and for full-screen PC games, nothing will do except to implement anti-aliasing over the full-screen (or scene as the technology is named).
HRAA
High-resolution anti-aliasing is basically the same thing as FSAA, except that HRAA is capable of eliminating that 'stair-case look' beyond 1024 x 768, into the high resolutions that we now demand :) Although HRAA is just getting started, anti-aliasing techniques are getting more robust at the same time that graphics processors are becoming exponentially more powerful. I predict that, in a couple years, anti-aliasing will be as common-place as textures - it will be a part of your graphics card and games without mention. Yep, just another spec to list :)
32-bit color
There seems to be a lot of confusion over what 32-bit color offers over 24-bit color, and rightly so because they both are capable of 16 million colors (16,777,216 to be precise) or 2^24th power. 32-bit hardware support was also something that was hyped at the time without much explanation, such as that NVidia's TNT2 could do 32-bit color but not 3dfx's VooDoo3. So if the total number of colors doesn't change, what are the extra 8 bits for? Alpha blending. It's a value that conveniently takes up 8 bits and refers to how transparent the rendered object is. Everything from completely opaque (like solid walls) to somewhat translucent (like water and fog) to completely transparent (like clear glass) can be described using these 8 extra bits. 32-bit games can then use this alpha blending value to create superbly realistic lighting effects with semi-substantial substances like variable fog thickness and light refraction due to glass and water. First-person shooters were the first games to adopt this technology, but since then, other 3D (flight sims and action games) and even 2D games (such as Diablo II) are using 32-bit color.
RAM bus width
For those who know how to calculate bandwidth (number of bytes transferred per second), you know that RAM speed and RAM bus width are required. Without going into too much detail for those new to 'tech specs', the bus width is measured in bits and more are better. For example, 100MHz RAM on a 128-bit bus width doesn't sound all that impressive compared to 200MHz RAM on a 64-bit bus width, mostly because speed is considered first, but a quick calculation reveals that both RAM set-ups have the same bandwidth and thus the same performance. However, since bus width is a very difficult spec to find, the burden is on me and I have provided this information for you. Aren't I nice?
nanosecond Rating
This can be used to calculate the theoretical speed limit of RAM, if you ever decide to do some RAM overclocking on your graphics card :) For instance, a 10ns RAM rating means a speed limit of 100MHz, while a 7ns RAM rating means a speed limit of about 143MHz. This calculation is actually quite simple and fast with the standard calculator program included with Windows. Take the ns number, divide by 1 billion (9 zeros), and hit the "1/x" or inverse button and there's your number in Hz. Divide by 1 million to get the MHz number, but you should be able to look at the Hz number and do the MHz conversion in your head. Imagine what 1ns RAM could do: 1GHz! Woohoo, I can't wait!
 |
|
|
|
|
|
|
Epinions.com ID:
computernut
|
|
Location: Eagan, MN
Reviews written: 18
Trusted by: 17 members
About Me:
A programming student, hardware is my hobby and gaming is my passion.
|
|
|
© 1999-2009 Shopping.com, Inc.