Analysis: AGP versus PCI-e

Forget about Crossfire and SLI for just a moment and concentrate on the root cause of the new graphics revolution -- Intel’s PCI Express (PCI-e). When last we left the Accelerated Graphics Port (or “AGP”), also an Intel creation, it had reached technical specification 3.0 and was known to most of us as AGP 8X. This final successor to AGP 1X, 2X, and 4X was officially released at the end of 2002, promising a 2.1GB/s bandwidth.

Zowie! That’s large -- or at least large enough to last for about two and a half years. Then, with the impending doom presented by the high bandwidth demands of high definition video and the constant push of gaming, it looked like AGP was running out of steam. It looked that way because it was. Shortly after the release of AGP 8X, Intel began talking about its successor, the Peripheral Component Interconnect Express (PCI-e) interface.

Advertised Specifications

PCI-e was originally called 3GIO (3rd Generation I/O). Its presence first became real in 2004, and sincere in 2005. So sincere, in fact, that it will soon kill AGP. PCI-e’s one-way bandwidth checks in at roughly 4GB/s and, if you get it going on two directions (full duplex), it will speed along at 8GB/s. While that’s anywhere from roughly 2X to 4X AGP’s bandwidth, if you really want to be wowed, think about it in terms of PCI’s original 132MB/s bandwidth.

But what do those numbers mean? At their root, they’re no more than a statistical analysis of the optimum performance under best-case conditions. Unfortunately, quantifying the reality of any speed differences between these two interfaces is not necessarily simple. An AGP motherboard and a PCI-e motherboard, no matter how similar, can never be identical. Manufacturing tolerances alone dictate that there must be a difference between the two.Test bed

How do you start? Perhaps as I did, with an ASRock 939 Dual SATA2 motherboard. The 939 Dual features both an 8X AGP and a PCI-e port on board. No matter which graphics option you choose, all of the ancillary components remain the same, as do all of the settings. And because of the graphic solutions I selected, even the graphics drivers were the same. It’s as close as you can get to the perfect scenario despite the fact that the very interfaces themselves are handled differently.

What graphics solutions were those? I pulled up a pair of Diamond MultiMedia’s Viper Radeon X1600 Pro cards, each stocked with 256MB, one PCI-e and the other AGP. The RAMDACS and core clocks for the two are the same, according to Diamond’s published specs, and they both share 12 Pixel Pipelines Architecture and an 8 Vertex Shader. These are both mid-range graphics cards with MSRPs of $200 (PCI-e) and $230 (AGP) although street pricing is considerably less.

The software of choice was COSBI Open SourceMark and FutureMark’s 3Dmark06. No effort was made to optimize any settings -- for either cards or the tests -- and the graphics resolution was set at 1024x768 and 32-bits.

<>And the Winner is…Both cards produced absolutely brilliant graphics and the ad hoc “testing” I did with video (I watched the wide screen version of King Kong -- hey, somebody's got to do it) was equally top notch. There was no discernable difference in quality between the two boards and no video hiccupping at all.

Graphics speed was something else, however. The PCI-e version of the Diamond Radeon X1600 Pro was noticeably faster running the 3Dmark06 tests and scored a 2,341 test result against the AGP’s 1,647. That’s a good gap -- about a 42 percent higher score -- but nowhere near the 400 to 800 percent improvement the specification claims. That differential should increase as you raise the resolution, but there’s no indication that, even then, it would ever come close to what’s advertised.

On the OSMark front, the PCI-e X1600 ran up a 1,526 score while its AGP sibling topped off at 1,334. That’s a 14 percent improvement for PCI-e, in this case, and there’s a reason for that. OSMark is a system-wide test with generic graphics segments. It’s geared not for the rigors of gaming but day-to-day Windows variety graphics.

What You Should Think Now

The easy conclusion is that PCI-e is certainly faster than AGP. Well, that’s a no-brainer. It was supposed to be. But should you switch to PCI-e from AGP? With Diamond’s X1600 selling for as low as $119 in some venues that too may seem like a no-brainer. But wait! There’s more! Unless you have a motherboard similar to my ASRock 939 Dual SATA2, you’ll also need a new motherboard that has a PCI-e graphics slot on it. That adds to the cost. Your new motherboard may not be compatible with your old CPU. And you’d better find out if your old memory modules will work with your new motherboard and CPU, too.As you might have detected by now, the move from AGP to PCI-e isn’t a simple matter of pulling out one card and slipping in another. It’s a weighty decision that can hit your wallet hard. The obvious questions are: 1) is a 40+ percent increase in gaming capability worth the expense, or; 2) if you’re not gaming, a 14+ percent increase in your daily computing?

The balance to those questions is that AGP is a dead end. You’re not going to find high-end graphics cards arriving in AGP format. To that extent, the answer is obvious: You will need to make the change at some point. With Vista on the way to redefining what your computer hardware should look like, with Crossfire arriving, and with SLI shaping up, it may not be a decision you want to make right now. It may be bad for the computer business but the most obvious indicator right now is to maintain the status quo for the next six to eight months if you’re not in a critical situation.