AFAICR from back in the day, nature was good for finding out if you cooling was inadequate, especially if you pre-warmed it with the full test. If you can go cold cold boot to pass nature only, it's probably heat. If it fails still doesn't rule it out, but it must be so bad it's crashing in games too.

BitWrangler wrote on 2023-05-30, 23:11:

AFAICR from back in the day, nature was good for finding out if you cooling was inadequate, especially if you pre-warmed it with the full test. If you can go cold cold boot to pass nature only, it's probably heat. If it fails still doesn't rule it out, but it must be so bad it's crashing in games too.

How many times do I need to say that it will pass all 4 games including nature flawlessly for multiple passes with driver 66.93.

Furthermore, with 45.23, it bombs out before rendering a single frame of nature everytime. It's not just like crashing randomly in the demo

Just out of curiosity, unlock he overclocking menu and turn the memory speed way down. Then see if it still crashes.

It could be bad caps that are causing problems with the memory. Here is evidence:

Repair shops hate me. I fixed defective memory on my video card with this one neat trick.

maxtherabbit wrote on 2023-05-30, 03:45:

With a FX5200 in windows XP, 45.23 crashes in 3dmark2001 on game 4, but 66.93 completes them all without issue. Haven't yet tried the same versions on 9x to see if the results are the same

I get that now, what we meant about 45.23 is focused on windows 98 games for wider games compatibility.

XP cannot always run games written for windows 98 and this 45.23 driver was written for window 98, keep in mind, which is partially true what you are seeing.

The best fit: Run cranky games written for 98. Also when you have any older games that will not run on windows 8 and windows 10 but works well on windows XP then you have the reason of stating your case.

I had JCSC Photo shop Pro 5.x got removed in windows 10 right after windows 10 got major version updated. There was no virus in it, and it was a original program installed from original media because I bought box of Photo Shop Pro with goodies years prior. It was written for early 32bits software, I think was for windows 98 too.

Cheers,

45.23 is a DirectX 9 driver from 2003 so yeah that's definitely intended for XP. Maybe that Nature crash is an issue with the XP version of the driver. I don't know if I've ever tried it there.

That driver is just a nice one for some old games, like some Lithtech 2.x games IIRC. If you don't find problems with newer drivers there's no reason to use it. You can also try 44.03 and 43.45.

I didn't have any problems running Nature or any other Windows 98 3DMark on a Pentium 3 with an FX 5700 or Quadro FX 3000 using driver version version 56.64.

There is a video from Phils (https://www.youtube.com/watch?v=HRhm4aGNI3o) that shows that version 56.64 is the fastest FX driver overall for DirectX, although 53.04 is not tested. 56.64 has an incompatibility with Thief 2 that can be trivially fixed with an INI setting. Everything else I have tried works perfectly. There is allegedly an incompatibility with the original Need for Speed that I have not tested; I haven't heard of any other issues.

The idea that only driver 45.23 is acceptable seems like a myth. I like 56.64 because it supports every FX series card, although there apparently is an issue on Pentium 2s: Re: GeForce FX driver testing on an Intel 440EX summary and report

I understand there's some controversy about driver cheats especially for the FX series, but it seems like these were also present in the 4x.xx drivers and mainly affected DX9 and 3dmark 03 specifically. Setting the image rendering to "quality" may disable some of them too. NVidia 4 series and FX generally has better texture filtering than ATI, and the FX has better AA than both as well.

With the FX and DirectX9 I can run Far Cry at 25fps on my Tualatin/Quadro box. I haven't tried nglide.

The FX series is killer in OpenGL, too.

copper wrote on 2023-06-03, 11:40:

I didn't have any problems running Nature or any other Windows 98 3DMark on a Pentium 3 with an FX 5700 or Quadro FX 3000 using […]
Show full quote

I didn't have any problems running Nature or any other Windows 98 3DMark on a Pentium 3 with an FX 5700 or Quadro FX 3000 using driver version version 56.64.

There is a video from Phils (https://www.youtube.com/watch?v=HRhm4aGNI3o) that shows that version 56.64 is the fastest FX driver overall for DirectX, although 53.04 is not tested. 56.64 has an incompatibility with Thief 2 that can be trivially fixed with an INI setting. Everything else I have tried works perfectly. There is allegedly an incompatibility with the original Need for Speed that I have not tested; I haven't heard of any other issues.

The idea that only driver 45.23 is acceptable seems like a myth. I like 56.64 because it supports every FX series card, although there apparently is an issue on Pentium 2s: Re: GeForce FX driver testing on an Intel 440EX summary and report

DirectX8.png

With the FX and DirectX9 I can run Far Cry at 25fps on my Tualatin/Quadro box. I haven't tried nglide.

The FX series is killer in OpenGL, too.

OpenGL.png

If the driver requires SSE it's a nonstarter for me

I think once upon a time I determined 56.64 is the last that works correctly with 440BX. At least with my Abit BF6 and whatever card I was using at the time (probably some FX series card). But I am also pretty sure that it breaks some old games and that's what got me experimenting with older drivers ages ago.

maxtherabbit wrote on 2023-06-03, 15:51:

copper wrote on 2023-06-03, 11:40:

I didn't have any problems running Nature or any other Windows 98 3DMark on a Pentium 3 with an FX 5700 or Quadro FX 3000 using […]
Show full quote

I didn't have any problems running Nature or any other Windows 98 3DMark on a Pentium 3 with an FX 5700 or Quadro FX 3000 using driver version version 56.64.

There is a video from Phils (https://www.youtube.com/watch?v=HRhm4aGNI3o) that shows that version 56.64 is the fastest FX driver overall for DirectX, although 53.04 is not tested. 56.64 has an incompatibility with Thief 2 that can be trivially fixed with an INI setting. Everything else I have tried works perfectly. There is allegedly an incompatibility with the original Need for Speed that I have not tested; I haven't heard of any other issues.

The idea that only driver 45.23 is acceptable seems like a myth. I like 56.64 because it supports every FX series card, although there apparently is an issue on Pentium 2s: Re: GeForce FX driver testing on an Intel 440EX summary and report

DirectX8.png

With the FX and DirectX9 I can run Far Cry at 25fps on my Tualatin/Quadro box. I haven't tried nglide.

The FX series is killer in OpenGL, too.

OpenGL.png

If the driver requires SSE it's a nonstarter for me

What? Your XP mindset do I assume? XP requires SSE and SSE2 with SP2 and SP3. The range where PII/ PIII of range 233 MHz to 800MHz is too underpowered for made to run in windows 98 era games and too fast for DOS games. Most of people went with PIII 900 through 1.2 and 1.4GHz so driver that requires SSE so this is not a issue either. Even needed more power went with P4 and AMD athlons processors, means does have SSE.

Geforce FX are too powerful and does not make sense to run on underpowered CPUs, and this does not matter as they certainly will have SSE support for CPU that is sufficient to power geforce4 and Geforce FX cards.

That why I'm questioning your mindset.

Cheers,

> Geforce FX are too powerful and does not make sense to run on underpowered CPUs

Well it really depends. High levels of AA/AF are not cpu dependent and some people pair FX5900U on earlier systems precisely so they can run earlier games at high res with max AA and excellent image quality.

pentiumspeed wrote on 2023-06-03, 20:57:

Geforce FX are too powerful and does not make sense to run on underpowered CPUs.

I have to agree with this. It really depends on your CPU as to whether there is any benefit to which GPU to use. I know that this quote suggests the entire series, but I would argue that the GeForce FX 5200 (128-bit) to be an excellent choice in a budget system. It's as good or better than a high end DirectX 7 GPU with the added benefit of shader features.

There is nothing wrong with a GeForce 2 MX or GeForce 4 MX, but an FX 5200 is also a good alternative.

I wish that the FX 5200 had performance equal to or better than the Ti 4200. I feel like the FX series would have been a lot more interesting if performance had scaled this way.

pentiumspeed wrote on 2023-06-03, 20:57:

..
XP requires SSE and SSE2 with SP2 and SP3.
...

This is definitely not true. Windows XP supports SSE/SSE2 CPUs out of the box, but even XP SP3 does NOT require even SSE(1).
Maybe it's because on AMD's side even the Thunderbird Athlons do not have full SSE support (so you can have a rather powerful 1.4 GHz CPU without SSE support).
AMD Athlon XP (Palomino) is the 1st generation from AMD that has full SSE support.
Both SSE and SSE2 requirements appeared much later in the Win XP POS updates.
Date of last update for XP Pro on non-SSE machines

pentiumspeed wrote on 2023-06-03, 20:57:

XP requires SSE and SSE2 with SP2 and SP3.

That's 100% not correct. XP works with pre-SSE (and even pre-MMX) processors on all service packs. Even with all officially released MS Windows updates. *Some* of the very late POSReady 2009 windows updates do require SSE, but they have been enumerated over on msfn.org and can be avoided easily enough.

pentiumspeed wrote on 2023-06-03, 20:57:

The range where PII/ PIII of range 233 MHz to 800MHz is too underpowered for made to run in windows 98 era games and too fast for DOS games.

"Too fast for DOS games" is irrelevant for almost all mainstream DOS games released in the VGA/SVGA era (with the exception of a handful of Turbo Pascal based titles which can be easily patched). I'm not interested in playing stuff written for XTs or CGA.

The part about 800MHz being too slow for Windows 98 strikes me as equally absurd but that's becoming quite tangential to the point here. My mindset is to build configurations that are interesting to me, like running XP on dual pentium pros. I don't give a shit about period correctness.

pentiumspeed wrote on 2023-06-03, 20:57:

Geforce FX are too powerful and does not make sense to run on underpowered CPUs,

FX5200 are cheap as dirt and similarly common. They offer reasonable performance on just about anything for little to no cost. It makes perfect sense to me

maxtherabbit wrote on 2023-06-04, 15:10:

The part about 800MHz being too slow for Windows 98 strikes me as equally absurd but that's becoming quite tangential to the point here.

It depends on when one sets the cutoff point for Win9x. Even if we use the year 2000 for that, there are still some games from that year (e.g. Deus Ex) which won't run well on such a CPU. But that's an edge case, and for the vast majority of Win9x games, something like an 800 MHz Pentium 3 will be more than enough.

FX5200 are cheap as dirt and similarly common. They offer reasonable performance on just about anything for little to no cost. It makes perfect sense to me.

Out of curiosity, does the FX 5200 suffer from text corruption in certain games, like the higher tier 5700 and 5900 cards?

maxtherabbit wrote on 2023-06-04, 15:10:

pentiumspeed wrote on 2023-06-03, 20:57:

The range where PII/ PIII of range 233 MHz to 800MHz is too underpowered for made to run in windows 98 era games and too fast for DOS games.

"Too fast for DOS games" is irrelevant for almost all mainstream DOS games released in the VGA/SVGA era (with the exception of a handful of Turbo Pascal based titles which can be easily patched). I'm not interested in playing stuff written for XTs or CGA.

Mostly agree on this. However, there are still speed-sensitive software (including Windows ones which often gets neglected) that aren't trivial to run with P6-based CPUs by simply using Throttle and/or disabling the L1 cache.

The part about 800MHz being too slow for Windows 98 strikes me as equally absurd but that's becoming quite tangential to the point here. My mindset is to build configurations that are interesting to me, like running XP on dual pentium pros. I don't give a shit about period correctness.

Period-correctness or not, that statement about 800 MHz being too slow looks elitist to me. I'm currently Windows 98 SE with a PIII-750 CPU just fine. 😀

Gmlb256 wrote on 2023-06-04, 16:19:

Mostly agree on this. However, there are still speed-sensitive software (including Windows ones which often gets neglected) that […]
Show full quote

maxtherabbit wrote on 2023-06-04, 15:10:

pentiumspeed wrote on 2023-06-03, 20:57:

The range where PII/ PIII of range 233 MHz to 800MHz is too underpowered for made to run in windows 98 era games and too fast for DOS games.

"Too fast for DOS games" is irrelevant for almost all mainstream DOS games released in the VGA/SVGA era (with the exception of a handful of Turbo Pascal based titles which can be easily patched). I'm not interested in playing stuff written for XTs or CGA.

Mostly agree on this. However, there are still speed-sensitive software (including Windows ones which often gets neglected) that aren't trivial to run with P6-based CPUs by simply using Throttle and/or disabling the L1 cache.

The part about 800MHz being too slow for Windows 98 strikes me as equally absurd but that's becoming quite tangential to the point here. My mindset is to build configurations that are interesting to me, like running XP on dual pentium pros. I don't give a shit about period correctness.

Period-correctness or not, that statement about 800 MHz being too slow looks elitist to me. I'm currently Windows 98 SE with a PIII-750 CPU just fine. 😀

Not only elitist it sounds crazy. As most of you guys should now by the many threads I have already made that i am also about the get my first win 98 gaming machine. During my research process I also checked the specs of he late 90s and early 2000 games I am in and for most of the games a 800mhz cpu would be seen as optimum. cpu for many games well into the year 2000. The statement that 800mhz is to slow for win 98 does not make any sense at all. At least not for most standard applications.

Joseph_Joestar wrote on 2023-06-04, 16:05:

Out of curiosity, does the FX 5200 suffer from text corruption in certain games, like the higher tier 5700 and 5900 cards?

I'm pretty sure that this bug is endemic to all GeForce cards (anything released after the TNT2).

See here:

Mipmap settings that fix Incoming (DirectX 5) on the GeForce FX

maxtherabbit wrote on 2023-06-04, 15:10:

FX5200 are cheap as dirt and similarly common. They offer reasonable performance on just about anything for little to no cost. It makes perfect sense to me

They are also fanless, which is a plus in my book, as well as commonly found with DVI.

Joseph_Joestar wrote on 2023-06-04, 16:05:

Out of curiosity, does the FX 5200 suffer from text corruption in certain games, like the higher tier 5700 and 5900 cards?

Yes, they have the texel alignment problem and lack the option to adjust it.

predator_085 wrote on 2023-06-04, 16:30:

Not only elitist it sounds crazy.

Way to insult posters who have been helping you with your myriad of questions. Just because some people prefer to play with non-period correct hardware doesn't make them "elitist" or "crazy".

For reference, there were always a couple of demanding games which needed more power than what was available at the time. As for the recommended specs, let's just say that some developers were overly optimistic with their estimates. Either that, or they didn't consider the desire for 60+ FPS in their calculations. But like I said before, such demanding games (like Deus Ex) are the exception, not the norm.