Shponglefan wrote on 2024-06-11, 01:32:

Main thing I'm trying to determine is games that would require 16-bit colors but wouldn't perform well on a PCI 3D card (e.g. Voodoo3 PCI or FX5500). I figure it's that time from about 1998-2000(ish) where such games likely lie.

Earlier 3D games would probably be fine with a PCI 3D card.

Depends on the resolution and frame rate that you're targeting. Some games from 1999 and 2000 can be surprisingly demanding, both on the CPU and GPU side, if high resolutions with AA+AF are desired. For example, I was testing Thief 2 at 1600x1200 with 4xAA + 16xAF and it just manages to stay above 60 FPS on my X800 XT card. This was in a huge, outdoor area covered with fog, with many enemies running around, so probably the worst case scenario.

Some other (not 16-bit color) games from that time are similarly demanding. In Deus Ex, it's difficult to get a locked 60 FPS in outdoor areas like Liberty Island (first level), thought that's mostly due to its insane CPU requirements. Drakan can also be quite demanding on both the GPU and CPU at max settings and high resolutions. And in Quake 3, the frame rate collapses whenever you approach a portal or a mirror.

Quake3 does run in 16-bit color but has a lot of benefit from 32-bit color (especially regarding overbrights and blended textures) so it's questionable to consider that a "16-bit color game".

Operation Flashpoint does 16-bit color, Glide and is also a cpu killer. The game was in development/publisher heck so there's a bit of legacy reach.

King's Quest VIII and Ultima IX are the only 98-99 case that's popping out in my mind for being excruciatingly slow.

I'd say all games *need* dithering for 16-bit color and there's some exceptions that force dithering off and they're japanese-developed playstation ports for some reason. If you worry the most about dithering appearance, consider PowerVR - they render internally in 24-bit and bring that dithered to a 16-bit buffer so the artifacts of overdrawn dither aren't there. Later Intel chips would do the same. There is no filtered 16bpp 'its 22-bit' nonsense here.

Joseph_Joestar wrote on 2024-06-11, 01:47:

Depends on the resolution and frame rate that you're targeting. Some games from 1999 and 2000 can be surprisingly demanding, both on the CPU and GPU side, if high resolutions with AA+AF are desired. For example, I was testing Thief 2 at 1600x1200 with 4xAA + 16xAF and it just manages to stay above 60 FPS on my X800 XT card. This was in a huge, outdoor area covered with fog, with many enemies running around, so probably the worst case scenario.

Some other (not 16-bit color) games from that time are similarly demanding. In Deus Ex, it's difficult to get a locked 60 FPS in outdoor areas like Liberty Island (first level), thought that's mostly due to its insane CPU requirements. Drakan can also be quite demanding on both the GPU and CPU at max settings and high resolutions. And in Quake 3, the frame rate collapses whenever you approach a portal or a mirror.

CPU speed should be fine if this ends up being either a Pentium 4 or Core2 setup. GPU is the challenge given that gap between the 3850 AGP versus a PCI video card (esp something like a Voodoo3).

Insofar as resolution that will depend on the monitor I imagine. The most convenient setup would be to use an LCD monitor with dual inputs (DVI or HDMI, and VGA). This way both cards could be routed to the same monitor.

Depending on the resolution of said monitor (likely either 1600x1200 or 1280x1024) would dictate performance targets.

leileilol wrote on 2024-06-11, 05:03:

I'd say all games *need* dithering for 16-bit color and there's some exceptions that force dithering off and they're japanese-developed playstation ports for some reason. If you worry the most about dithering appearance, consider PowerVR - they render internally in 24-bit and bring that dithered to a 16-bit buffer so the artifacts of overdrawn dither aren't there. Later Intel chips would do the same. There is no filtered 16bpp 'its 22-bit' nonsense here.

I have been considering trying out a PowerVR PCX2 in the mix. I think a triple or even quad video card setup might be possible, just to see how many bases can be covered all on one system.

Another update, I recently acquired an Axiom Tech IMB200 motherboard. It's similar to the DFI motherboard I'm currently using, but the IMB200 has an extra PCI slot.

I've currently got it on my testbench as I try it out and make sure everything is working. I've tested it with an Orpheus II LT ISA sound card and DOS audio works perfectly with it.

I may look to this new board to replace the current DFI motherboard. This would allow me to add the AWE64 without needing to change the existing card setup. If I figure out a single-slot cooling solution for the GeForce4, I could even add another card like a second Voodoo2 for SLI capability.

Shponglefan wrote on 2024-06-12, 20:35:

Another update, I recently acquired an Axiom Tech IMB200 motherboard. It's similar to the DFI motherboard I'm currently using, b […]
Show full quote

Another update, I recently acquired an Axiom Tech IMB200 motherboard. It's similar to the DFI motherboard I'm currently using, but the IMB200 has an extra PCI slot.

AXIOM Tech IMB200 motherboard.jpg

I've currently got it on my testbench as I try it out and make sure everything is working. I've tested it with an Orpheus II LT ISA sound card and DOS audio works perfectly with it.

I may look to this new board to replace the current DFI motherboard. This would allow me to add the AWE64 without needing to change the existing card setup. If I figure out a single-slot cooling solution for the GeForce4, I could even add another card like a second Voodoo2 for SLI capability.

AXIOM Tech IMB200 testing 3.jpg

sounds like a good plan, that board seems to have been designed for a project like this 😀

Looks like an upgrade! I think your older board used a Winbond pci-isa bridge? So this should have even better compability if the ITE IT8888 on this board is as good as they say!

As these are Intel 865p boards do they allow you set down to or below 100MHz FSB (400Mhz Quad Pumped) like 478pin variants? So you can half the speed of a 200MHz FSB CPU and then get more speed options.

A tool called CPUCOOL and the smaller CPUFSB worked on a P3 Xeon board I had which didn't have full bios fsb options really well.
So you can boot to W98 -> CPUCOOL set fsb xxx mhz -> reboot and it sticks till power down.

gerry wrote on 2024-06-13, 08:12:

sounds like a good plan, that board seems to have been designed for a project like this 😀

It does indeed! As long as it throttles the same way as my current board, it should be good to go. 😁

Hanamichi wrote on 2024-06-13, 16:44:

Looks like an upgrade! I think your older board used a Winbond pci-isa bridge? So this should have even better compability if th […]
Show full quote

Looks like an upgrade! I think your older board used a Winbond pci-isa bridge? So this should have even better compability if the ITE IT8888 on this board is as good as they say!

As these are Intel 865p boards do they allow you set down to or below 100MHz FSB (400Mhz Quad Pumped) like 478pin variants? So you can half the speed of a 200MHz FSB CPU and then get more speed options.

A tool called CPUCOOL and the smaller CPUFSB worked on a P3 Xeon board I had which didn't have full bios fsb options really well.
So you can boot to W98 -> CPUCOOL set fsb xxx mhz -> reboot and it sticks till power down.

I haven't tried throttling the FSB yet. There are no BIOS options on these boards unfortunately. I've looked into a few tools, but the other issue is PLL chip support. I'll have to see which tools could potentially work.

Good to know that it can be set under Windows and retained on soft reboot. DOS options for setting FSB seem quite limited.

Started trying some FSB tools to see if it might be possible to change the FSB on the DFI G7S620-N motherboard.

The DFI motherboard uses the ICS 952617 PLL chip.

Unfortunately it doesn't seem any FSB tool directly support this chip. I decided to try a couple of them anyway to see what would happen.

SetFSB has a couple similar options (ICS952611 and ICS952619). Trying them out simply resulted in the system freezing. Even after a hard boot, would result in the system failing to boot and required a full power off before booting up again.

I then tried CPUCool. It also froze, however upon rebooting it complete corrupted my Windows XP install. Even trying a system recovery didn't work and I ended up having to do a complete reinstall.

Unfortunately I did this with my main working build rather than my test bench. In hindsight I should have set up a test install of XP rather than risking my main system.

At any rate, I'm going to park testing FSB tools with the DFI motherboard and perhaps test out the IMB200 instead.

I've been testing out the AWE64 and Orpheus II combo (using the LT as a stand-in).

Since the Orpheus II and AWE64 each use a pair DMA channels, I'll need to activate only a single card at a time. I don't foresee this being issue since I can't think of a scenario where I would want both cards running at the same time. The exception would be for MPU-401 support on the Orpheus II, but since that is hardware-only (no driver), it still works even with the Orpheus II disabled via drivers.

The trickier thing will be figuring out CD audio. I can imagine scenarios where I would want CD music with both cards. Not sure if I can just split a CD-audio cable and route it to both cards? Or if there is a another solution for that?

Shponglefan wrote on 2024-06-23, 19:27:

The trickier thing will be figuring out CD audio. I can imagine scenarios where I would want CD music with both cards. Not sure if I can just split a CD-audio cable and route it to both cards? Or if there is a another solution for that?

Not sure if the Orpheus has a CD_SPDIF input, but if so, you can use a 2-pin SPDIF cable from the CD-ROM drive on that card, and the regular (4-pin) analog cable for the AWE64. I used to do this for one of my rigs which had a SBLive and an ESS 1868F.

Alternatively, you can route the analog CD audio output to an external mixer and not connect it to any sound card. Might actually lower the noise a bit.

Joseph_Joestar wrote on 2024-06-24, 00:52:

Not sure if the Orpheus has a CD_SPDIF input, but if so, you can use a 2-pin SPDIF cable from the CD-ROM drive on that card, and the regular (4-pin) analog cable for the AWE64. I used to do this for one of my rigs which had a SBLive and an ESS 1868F.

Alternatively, you can route the analog CD audio output to an external mixer and not connect it to any sound card. Might actually lower the noise a bit.

Unfortunately the Orpheus cards don't have a 2-pin SPDIF input.

Hadn't considered just routing CD audio directly to a mixer. I'll have to give that one some thought and see how I might incorporate that.

This past weekend I migrated my 'studio' PC to a different case. This freed up the Antec case I had been using for that PC.

As I redo my Pentium 4 build, I may use this Antec case. It is bigger than the existing Define R5, but has an added 5.25" drive bay (3 total) versus the R5. The extra drive bay gives me the opportunity to add a CF adapter.

Standard cooling is a pair of 120mm front intake fans and a 120mm exhaust fan. The case also features a couple mounting points for internal 120mm fans. This would be useful for added cooling for the graphic cards, since the top fan would blow right across those slots.

I'm not 100% decided if I'll use this case yet, but I am leaning towards it. The addition of a CF slot would be nice for removal storage, and an added internal fan could help with cooling.

One thing I've always kind of wanted for this build was a more traditional beige case to match the rest of my intended setup.

While I was debating case options for this build this week, I stumbled across this case at a local thrift shop. I think it's an Antec SX 830 (?) and it fits that classic beige aesthetic.

The case does have some drawbacks for cooling compared to newer cases. It only has options for 80mm fan mounts versus newer cases that support 120mm or 140mm. Although this case does at least have dual fans for both intake and exhaust. However, it also lacks any other ventilation. I'm surprised it doesn't even have passive ventilation holes on the bottom or next to the expansion slots.

It also lacks cable routing options and has no front I/O ports.

Still, I do love that classic beige. 😁

If I decide to use this case, I'd forego plans to try to add a front mounted CF slot and likely just stick to the SSDs for storage.

On the other hand, I'm now considering going with a dual optical drives: DVD-ROM and CD-ROM. The latter could be a slower model (e.g. 6x or something) and mitigate the need for throttling software. Dual optical drives may also solve my CD-audio dilemma with the AWE64 and Orpheus II. I could run an audio cable to each sound card.

At any rate, I have to finish tearing down this system and then do some more planning to see which case I ultimately want to use.

That's one of the cases everybody wants! Congratulations on finding one!

H3nrik V! wrote on 2024-07-07, 22:58:

That's one of the cases everybody wants! Congratulations on finding one!

Thank you! It was a lucky find for sure! 😁

As I plan to migrate my build to the above Antec case, I tested a bunch of beige DVD drives I have on hand. Main thing was to see which drives worked, how noisy they are, and how compatible with tools like Nero DriveSpeed.

From top to bottom:

1. LG GDR-8161B
2. Pioneer DVR-110D
3. LG GSA-H50N
4. Lite-On XJ-HD166S
5. Sony DW-D22A
6. LG GDR-8161B (second one)
7. LG GSA-4040B

LG GDR-8161B

One of the LG GDR-8161B drives had disk read errors. Its opening/closing drive mechanism also seemed a bit clunky. I'll have to see if it can be repaired.

The other LG GDR-8161B worked. Nero DriveSpeed reported various options including 20x, 17x, 16x, 13x, 8x, and 3x. It was still a bit noisy at 8x, but silent at 3x.

LG GSA-H50N and Pioneer DVR-110D

The LG GSA-H50N and Pioneer DVR-110D failed to support Nero DriveSpeed. At full speed they were quite loud as would be expected.

Lite-On XJ-HD166S

The Lite-On XJ-HD166S had partial Nero DriveSpeed support. It didn't show any speed options when loading the program, but it did respond to custom speed inputs. However, at lower speeds it made loud clicking noises.

Sony DW-D22A

The Sony DW-D22A supported Nero DriveSpeed, but made loud clicking noises even at slower speeds.

LG GSA-4040B

The LG GSA-4040B supported a few different native speeds with Nero DriveSpeed including 32x, 24x, 16x and 10x. At 10x speed the drive was virtually silent, yet still worked fine with the testing I performed.

After all the testing, I think the LG GSA-4040B is my preferred choice. I'll have to do more thorough testing to see how it performs, but so far it seems reliable and quiet.

Recently acquired another AWE64 Gold,which I paired with a SIMMConn memory card. I'm going to look at adding it to this build in place of the previous AWE64 (Value) I was testing.

I still need to figure out hardware resource assignments.

With the AWE64 requiring two DMA channels (low/high), I may try reserving a total of three DMA channels for both the ISA sound cards. Between two IRQ and three DMA, I could in theory use both the Orpheus II (Crystal audio) and AWE64 at the same time.

Since I'm planning on dual optical drives, this means I could use CD-audio from one card and digital audio from the other. I don't know if there will be many scenarios where I necessarily require that, but having the option couldn't hurt.

The only thing I'm not sure about is if reserving three DMA channels will be an issue when it comes to any of other hardware and Windows installs. I suppose I could try this on my existing build and see if it causes any issues.

Did some initial testing of the AWE64. I wasn't sure which RAM would be compatible with it, since from what I've read SIMM compatibility can be hit-or-miss with the SIMMConn.

Fortunately it worked with the very first 32MB SIMM I tested. This particular stick of RAM uses Toshiba TC5117400CSJ-60 chips.