Wikipedia mentions "Dynamic Adaptive Speculative Pre-Processor", which allegedly acts like L3 cache.
In theory this should give Durons a notable boost. Will see if I can find my nForce board and maybe compare it to a different one with a Duron.
Yeah it does seem likely that the nForce chipsets have the best Socket A memory controller and it should help Duron out. I'm not sure if it would benefit Duron more than other CPUs though.
Was mATX really such a taboo thing in 2000 that they said let's just extend the board? I find this hilarious, I half expect to see markings for the end user to cut around the excess board whenever I look this board up again. 🤣
I've seen some intel boards like that too.
Basically, just take their mATX design and layout, add an extra PCI slot or 2, which is pretty easy in terms of routing, then manufacture.
I gave Call of Duty 1 a try for maybe the first time ever yesterday. I ran into some funky issues! Radeon 8500's mip mapping stops working after a few seconds. So you go from smooth texturing to what looks like Matrox Mystique texture filtering. Extreme texture aliasing. Apparently there's a bug with the chip or driver that causes this in some cases with some OGL pixel shaders. One net post says it's a fancy water effect sometimes present (link). Catalyst 5.8 and 3.10 behave the same so it seems like it was never addressed.
I switched to a FX 5200 to experiment and found that card suffered extreme slowdown after a few minutes. Sometimes with texture corruption. Tried a GF4 Ti with the same result. Tried going from driver 56.64 to 45.23 and this seems to fix that issue.
So I went ahead and did some (quite poor, admittedly) testing.
I compared an ECS K7S5A motherboard with the SiS735 chipset against MSI K7N420 Pro with the nForce420 chipset. Used a Radeon x800XT PE as the video card (catalyst 6.12), WinXP SP3, 1GB RAM.
Firstly, I wanted to test if the supposed "almost-l3-cache" Dynamic Adaptive Speculative Pre-Processor helped the poor little Duron compensate for its lack of L2.
Comparing the gains of going from Duron to Athlon, on the SiS745 the Athlon gained 15.25% in UT and 21.83% in Q3, while on nForce it gained 24.38% and 16.48%. So overall, no, the nForce did not help close the gap between Duron and Athlon.
Second, I wanted to test how the SoundStorm audio helped reduce CPU overhead and compared 3 cards - An Audigy2 ZS, SoundStorm integrated audio and a more CPU dependent CMI8738.
Overall, nvidias sound solution does seem to remove quite a bit of CPU overhead, which is quite nice.
Other things I noticed - the ECS motherboard did not like it when I used faster memory. I suspect this might be due to an odd memory-fsb ratio as the 1GHz Athlon performed quite well with the 133 option.
Audigy2 card did not like Half-Life. I'm not sure if this is due to driver overhead or something else, but that card crippled performance and was worse than the CMI card.
Lastly - dual channel memory adds pretty much nothing for gaming when it comes to these lower end Athlons.
I wish I could have done it a little bit better and more in-depth, like at least updating to the latest bios, but this took me way too much time as it did and this does provide some basic info.
Yeah the dual channel is mostly there for the IGP case.
Which Audigy drivers did you try?
I've had good luck with the NV APU in a lot of different games. EAX 1/2 and 3D audio work well. The OpenAL works too in older games that use it. Unreal 2 and SOF 2 recognize it. Doom3 won't take it though since I suppose it looks specifically for EAX 4.
AGP compliance/compatibility is suspect though. Did you check if the ATI drivers set you to AGP 1x? Both 8500 and 9800 get set that way on my board and I've seen old forum posts talking about it too.
I have no idea if different Audigy drivers might help. I always use the Daniel K Audigy Support Pack these days. Back in 2001-2003, Audigy drivers were not exactly rock solid. Then the effort to get EAX 4 into Doom3 caused a bit of a rewrite from what I've gathered and who knows what that did to older game compatibility. Doom3 doesn't work correctly either beyond some old beta driver. 🤣
Well I just tried the slightly older drivers and then some early ones from a CD on Vogonsdrivers and neither times could it break the 40 fps mark. SoundStorm is a very capable solution and I don't see much reason to use a dedicated card anyway.
Gpu-z reports that the x800 was indeed running in 1x mode.
I have the MSI K7N420 set up at the moment and tried to play some Max Payne on it. I found the game skipping frames during the intro and audio looping during play. It's a problem with the nForce audio on 98SE. Swapped to an Aureal card and the game worked correctly. Windows 2000 also worked fine with the nForce audio.
