Ydee wrote on 2023-09-10, 09:40:

Of course I can, here they are. I only have one exclamation mark there, and that's for the SB16 emulation, which unfortunately can't be activated because of the lack of chipset support, no support for DDMA or TDMA, only PC/PCI (missing header), LPC or F type DMA are supported.

Thanks! I am genuinely impressed. I will certainly have to experiment with later AMD platforms now... Does NUSB36e work with the USB2 ports or are you stuck with USB 1.0 speeds only?

Ydee wrote on 2023-09-10, 09:40:

Of course I can, here they are. I only have one exclamation mark there, and that's for the SB16 emulation, which unfortunately can't be activated because of the lack of chipset support, no support for DDMA or TDMA, only PC/PCI (missing header), LPC or F type DMA are supported.
<snip>

I was inspired by your build to do a quick test with my Asus M3A78-EM and I am glad to report that everything works very well indeed! Using "setup /p i", patchmem and ptchsata yields an almost perfect install, with only the SMBUS showing an exclamation mark in Device Manager.

In the meantime I tested with a 90nm Sempron @ 1.8Ghz. My Wolfdale @ 3Ghz scores around 515fps in my Q3A demo001 benchmark, I got around 350 with this setup. But I just purchased a 3Ghz 45nm part, which I will compare with the Wolfdale when it arrives. I'm very curious to see which is faster - in theory the Wolfdale should be better, because it has 6MB of L2 as opposed to 2MB for the dual-core Regor, but who knows.

The tradeoff between the Intel and AMD platforms then becomes TDMA/DDMA and DOS multiplier changing (Intel lacking any multiplier control, and AMD on AMD/ATI southbridge lacking TDMA/DDMA -- TDMA works on VIA PCIe platforms though, but you're stuck with ancient 2005-era K8 CPUs at best as opposed to 45nm 2010/2011 era CPUs with the 760g/780g AM2/AM3 platform). But like you say, I'm not interested in DOS with this machine, it's meant to be a fast Windows 98/Windows XP dual boot. I am leaning towards the AMD. The motherboard is more robust than the Intel.

mockingbird wrote on 2023-09-10, 16:36:

Thanks! I am genuinely impressed. I will certainly have to experiment with later AMD platforms now... Does NUSB36e work with the USB2 ports or are you stuck with USB 1.0 speeds only?

Let me answer my own question: Yes, it certainly does have a driver for the AMD USB 2.0. It needs to be manually installed in device manager after installing nusb36 - after the reboot.

EDIT: Well I may not have a 45nm Regor to test with yet, but I do have in my possession this 90nm Windsor 3Ghz beast (6000+). I have no idea where I got it from, but it works. I get 478.2 fps compared to 515fps with the Wolfdale. The question is will Regor close (or surpass) that 7% gap... We'll see.

EDIT (09/20/2023): I received my Regor CPU and tested it in Quake3... I ran it at its default of 2.9Ghz (I would not pay the premium for the 3Ghz variant) and got 495FPS... So I would put it on par with the Wolfdale (or maybe a tiny bit slower), but it is multiplier locked, and unlike any AMD chip I dealt with in the past in this regard. It also runs nice and cool compared to the older models with the larger lithography. I'm not sure if the heatspreader is soldered on or not (all the older ones aren't), but it runs cool enough that I'll leave it alone.

mockingbird wrote on 2023-09-11, 03:45:

I was inspired by your build to do a quick test with my Asus M3A78-EM and I am glad to report that everything works very well indeed! Using "setup /p i", patchmem and ptchsata yields an almost perfect install, with only the SMBUS showing an exclamation mark in Device Manager.

This, too, has a solution: install this version of the driver (catalog) for SMBus, which supports W9x systems and makes the exclamation point disappear.

I tried the W98SE on this board because it's uATX and it fits perfectly at a table in a small case and because I had it at my disposal along with CPU and memory for a very low price.
Ragnargd wrote on MSFN that on set AMD 770/710 can't start W98SE because it suffers from instability and other problems. So I thought I'd give it a try.
I'm not seeing any trouble yet, but maybe it's because it's an obscure hybrid with a socket 754 and a relatively modern (then!) AMD780/710 chipset. Obviously was s.754 paired with nForce 3 or 4, VIA K8T8xx or ULi 16xx, maybe ATI RS.

Ydee wrote on 2023-09-11, 13:44:

This, too, has a solution: install this version of the driver (catalog) for SMBus, which supports W9x systems and makes the exclamation point disappear.

Can you please check your post again? I do not see a link. Thanks

I tried the W98SE on this board because it's uATX and it fits perfectly at a table in a small case and because I had it at my disposal along with CPU and memory for a very low price.
Ragnargd wrote on MSFN that on set AMD 770/710 can't start W98SE because it suffers from instability and other problems. So I thought I'd give it a try.
I'm not seeing any trouble yet, but maybe it's because it's an obscure hybrid with a socket 754 and a relatively modern (then!) AMD780/710 chipset. Obviously was s.754 paired with nForce 3 or 4, VIA K8T8xx or ULi 16xx, maybe ATI RS.

Same here... I have to of these 780g boards sitting around for years now and was always wondering what to do with them... Do you know which RLOEW nvidia patch he is referring to in his post (besides the VRAM patch)?

mockingbird wrote on 2023-09-11, 14:03:

Can you please check your post again? I do not see a link. Thanks

IIRC attachments predate inline hypertext in email.

Ydee wrote on 2023-09-11, 13:44:

This, too, has a solution: install this version of the driver (catalog) for SMBus, which supports W9x systems and makes the exclamation point disappear.

Thanks! Everything is now installed and working great. This will serve as a tremendous asset for my 45nm speed comparison with Intel. (as soon as my CPU comes in).

mockingbird wrote on 2023-09-11, 14:03:

Do you know which RLOEW nvidia patch he is referring to in his post (besides the VRAM patch)?

You mean this?:

For NVidia-cards on 890 and AMD 9xxx chipsets, to avoid stuttering because of an incompatibility of the NVidia-Driver with those chipsets, you need a commercial patch by RLoew. It also includes the NVidia-Ram-Size Fix, so this gives you further options.

I think that's a package of patches from the NVVGAFIX directory of Rudi Loew's utilities package:

• CONTENTS
  
  PATCHNVC.EXE Graphics Driver Patch
  PTCHNVSZ.EXE BIOS Image Patch Program
  PATCHOPT.EXE Patch PCI.VXD to Disable Video Option ROM Config
  NVFLASH.EXE nVidia BIOS Flash Program (Copyright www.techpowerup.com)
  CWSDPMI.EXE DPMI Support Program (Copyright CW Sandmann)
  NVSIZE.EXE Display Video RAM and MMIO Space Sizes
  README.TXT This File
  MANUAL.TXT Users Manual
  LICENSE.TXT License terms and conditions
  HISTORY.TXT Revision History
  FIXINTR.VXD Interrupt Fixup VXD (All IRQs)
  FIXINTR#.VXD Interrupt Fixup VXD (Individual IRQs) #=IRQ
  FIXEOI.VXD Alternate Interrupt Fixup VXD (All IRQs)
  
  NVCHECK.EXE Compatibility Check Program (included in free Package)
  TESTEOI.VXD Test VXD for Interrupt EOI bug (included in free Package)
  NVCHECK.TXT Compatibility Check and Test Manual (included in free Package)

and maybe even Nvidia video driver shutdown fix.

I'm sorry about my English and the wrong line code, those are my weak points.

Ydee wrote on 2023-09-12, 15:27:

You mean this?:

For NVidia-cards on 890 and AMD 9xxx chipsets, to avoid stuttering because of an incompatibility of the NVidia-Driver with those chipsets, you need a commercial patch by RLoew. It also includes the NVidia-Ram-Size Fix, so this gives you further options.

Thanks... I wonder if the incompatibility (besides the 512mb+ VRAM issue) is referring to this in the readme:

A set of Interrupt Fixup Files (FIXINTR) are provided to handle loss of
Interrupts due to mishandling of End Of Interrupt (EOI) commands by the
NVidia Graphics Driver. Symptoms of this problem include Mouse Pointer
freezing, DISK Write Error BSODs and USB problems.

Seems it affects other chipsets too (saw a reference to z97).