Is a motherboard with only 64Mb Graphics Aperture Size a limitation?

[Trinity]

I'm considering the Asus CUSL2-C + PIII 1 GHz cpu to go with my Marvel G400. This board uses the 815EP chipset which limits the graphics aperature size to 64M vs 256M on my P3B-F. What impact if any does this have on video capture, editing, mpeg production?

I know Doc pointed out a problem with using the RT2000 for MPEG 2, but a subsequent poster indicated that this would be corrected by an updated software release from Matrox.

Given that I don't use my Marvel for games and don't intend to upgrade to the RT2000, will the 64M limitation be of any consequence?

[FrankDC]

If you don't play 3D games, no.

[Dr Mordrid]

Maybe, maybe not. It depends on what effects and software you use when editing.

A lot of programs and plugins use OpenGL and/or Direct3D/DirectDraw HAL (hardware abstraction layer) rendering in building their previews and interfaces at the very least.

Both of these features work faster & better with a large AGP apature because of the involvement of the display adapters OpenGL hardware, if present. The Marvel has it.

VizFx is an example of a plugin with OpenGL support. Check the bottom of this page for the OpenGL recommendation;

http://www.vizone.com/product/vizfxm...xmspsysreq.htm

Cool3D is an example of a program with a HAL rendering option you can turn on if your card supports it.

Others include HollywoodFX, BorisFX & Boris RED 1.5, DigiEffects (After Effects plugin), Real Motion (AE plugin) etc, etc.

Also many editors use 3D programs of one type or another ranging from shareware to trueSpace, Rhino, Lightwave and 3D Studio MAX.

The vast majority of these use OpenGL for their interfaces at the very least, which assists productivity a lot. You should try using a 3D program without it. Painful.

A large AGP apature is of considerable advantage for these programs as well, again because of the assist it gives hardware acceleration.

They aren't just for games, especially when the project gets complex.

Dr. Mordrid



[This message has been edited by Dr Mordrid (edited 03 February 2001).]

[Trinity]

Thanks, Dr. Mordrid.

Don't you just hate it when products are deliberately crippled for business reasons rather than due to technology?

When you say this affects the use of OPEN GL, do you mean that things will simply run slower, or will there be things this board just cannot do because of the 64Mb limit?

[FrankDC]

Good grief, thanks for the info. I had no idea AGP aperture is used so extensively outside the 3D gaming world.

What was Intel thinking when they put the 64MB limitation on the 815??

[Trinity]

IMO, the 815 chipset is only intended as a stop-gap to draw users back that were put off by costs associated with upgrading to the 820. If I switch to this motherboard, it will be the first time that an 'upgrade' will result in actually losing functionality from my previous board. The most glaring deficiencies with the 815 vs my 440BX is the 512M memory limitation and the 64M aperture. I believe Intel's goal was to make a chipset good enough for 440BX owners to upgrade, but not so good that it would steal business away from their 820s...a half-hearted attempt at the perfect board. On-board video and sound are usually tip-offs to a low-end board. Luckily, Asus has the cheaper CUSL2-C with the crippled video so that you don't have to pay more for something you'll never use.

If you want a P3 mobo that is 133FSB that isn't VIA or AMD,I think the CUSL2 (or CUSL2-C) is one of the best. I'm a bit of a tweaker, so the ability to overclock in 1 Mhz increments is very attractive to me. I think it will suffice until my next upgrade to 2 Ghz and beyond.

[DosFreak]

If you need a dual mobo the VP6 is a suprisingly stable board. I'm impressed with it. Have all AGP options and everything tweaked to the Max in the bios. No crashing or strange problems at all in Windows 2000 Pro.

I'm happy.

[Dr Mordrid]

Most times it affects the throughput of the AGP system. To the extent that this slows down the transfer of data in and out of the graphics card it can cause enough of a slowdown to affect the speed of user interfaces and preview rendering, especially with realtime effects, OpenGL or even Direc3D renderings that use HAL.

Fortunately it looks like the RT-2000 will not be so impared for long. The guys found a way around the problem for its future driver updates, but the issue still applies to the other software I mentioned.

This mostly applies when the project is large enough to hit the bottlneck. Just twirling a polygon around in trueSpace won't cause a problem. Working with a more detailed model or a complex Cool3D project you can see a larger effect.

As to what Intel was thinking...who knows? They would have served everyone a lot better if they had just updated the 440BX microcode to include 133mhz, DDR, ATA100 and other new features, but then that might have taken some marketing thunder out of the P4. Not to mention their abortive arrangement with RAMBUS. Yecchhh....

My bet is that with such an updated 440BX setup and a fast PIII the P4/RAMBUS would have likely at least met its match.

The uppance? With the 440BX Intel produced a consumer chip with professional speed. Marketing probably had a stroke....

Dr. Mordrid


[This message has been edited by Dr Mordrid (edited 04 February 2001).]

[FrankDC]

Well, the various chipset and DDR memory fiascos will probably cause lots of people (me included) to simply upgrade their BX systems with faster P3's instead of migrating to P4. Even with DDR the P4's benchmarks have been completely underwhelming.

And speaking of Rambus, I wish they'd take a long walk off a short pier. The way they're going around suing companies, you'd think they invented the concept of clock doubling. Apparently their lawyers have never heard of SCSI-2 or AGP..

[This message has been edited by FrankDC (edited 04 February 2001).]