VOGONS


First post, by tenyuhuang

User metadata
Rank Newbie
Rank
Newbie

While building my own Windows 98 machine, I took a scenic route and benchmarked my Ti4200 against various versions of Nvidia driver.
It seems that performance DO get worse as the driver get newer and newer as opposed to a speculation where they could get better with a certain update.
I didn't benchmark EVERY single version of the driver, so I might be missing the gap; I'll be glad to know if the few versions I missed actually had performance boost.

Below is a detailed datasheet of the benchmarks; as a reference, I've included a 3DB result browser database in the attachment.

R0zlJOk.png
8xPK1kw.png

A few notes regarding this test:
1) All benchmarks were ran twice, and the better of two was kept individually;
2) The Ti4200-8x runs at AGP2x, on VIA 4-in-1 4.35;
3) System main bus runs at 133MHz, with PCI clocked at 1/4 ratio;
4) The OS is a clean install of Windows 98 SE without any updates except Direct X 9.0c;
5) Power is rated at 350W - shouldn't be a problem.

There always seems to be this mystery around Nvidia drivers, and there doesn't seem to be a certain conclusion, either.
I hope this little test result could be of help.

Of course, there could be some human errors introduced due my lack of experience (even though I didn't change any hardware or software except for the driver); that said, if you noticed anything wrong in the database (which included my test system and environment), I'm eager to know 😊

Attachments

  • Filename
    Ti4200 Driver Benchmark Database.zip
    File size
    23.56 KiB
    Downloads
    62 downloads
    File comment
    3DMark 2001SE Result Browser Database
    File license
    Fair use/fair dealing exception

Reply 1 of 2, by FFXIhealer

User metadata
Rank Oldbie
Rank
Oldbie

Man.... that is a lot of benchmarks. Must have taken you a while.

What I'd like to see - and it's probably been done before - is tests showing which graphics cards are bottlenecked by different AGP speeds. Like at what point does AGP 2x start bottlenecking an AGP 4x card? Ya know what I mean? Like, I have an AGP 4x motherboard for Windows XP with an Athlon XP Barton. So I can't use 8x only AGP cards and a lot of 4x/8x cards... are they hampered by the lack of 8x speed available?

Like, we all know that current-gen graphics cards like the 1070 don't fully saturate the PCI-Express 3.0 16x bus speeds yet. The 1080 and Titan XP come damned close I'd imagine. But these cards all still work pretty well in a PCI-Express 2.0 slot too.

292dps.png
3smzsb.png
0fvil8.png
lhbar1.png

Reply 2 of 2, by tenyuhuang

User metadata
Rank Newbie
Rank
Newbie
FFXIhealer wrote:

What I'd like to see - and it's probably been done before - is tests showing which graphics cards are bottlenecked by different AGP speeds. Like at what point does AGP 2x start bottlenecking an AGP 4x card? Ya know what I mean? Like, I have an AGP 4x motherboard for Windows XP with an Athlon XP Barton. So I can't use 8x only AGP cards and a lot of 4x/8x cards... are they hampered by the lack of 8x speed available?

That was almost my original intention. I was planning to do an AGP 2x vs 4x comparison since I got myself a VIA 694x, why not abuse it out of some nice 4x performance? 😊
But in the end I failed miserably trying to get this stubborn VIA thing to work; to my surprise, that 694x seriously isn't happy - it doesn't want lunch; it wants breakfast:

intOU6o.png
Ocnm5CJ.png

*All benchmarks are made in AGP 2x.

I suppose I've got myself a bad mobo. So I might wanna go plan B, but at current state I'm out of budget and in a cooldown.
Will definitely do a 2x vs 4x vs 8x test as soon as I got myself a worthless (not really) Pentium 4 platform 😊

PS: But I can definitely tell you that PCI is absolutely going to bottleneck this Ti4200; I accidentally got it to work under PCI mode and it got a ~20% performance nerf.