hydraphone wrote on 2025-12-02, 15:41:

... AGP has it's own "lane" that is separate from the PCI.

It is not clear to me whether all PCI slots share the single 32-bit 33MHz interface or not, but if they do, that makes it clear that I need an AGP card for this build.

Again: AGP is another kind of parallel PCI bus, so we do not talk about serial lanes from PCIe. AGP bus and PCI bus are bridged by 440LX chipset. That means a PCI TV card can write into video memory of an AGP card without blocking the CPU for transfers.

Yes, all PCI slots use parallel address and data lines. However, some wires like the controls for INTA, INTB, INTC and INTD or the enumeration for PCI slots are different.
To not slow down the PCI bus and to achieve a faster transfer rate AGP bus had been invented. The Intel i740 graphics card also demonstrates how to deal with textures remaining in main memory due to insufficient amount of graphics RAM. Texture loading would have been a pain on PCI.

Disruptor, no need to be patronizing --I already put the word lane in quotes after seeing pages #2 and #10 of https://bitsavers.org/components/intel/_dataS … 43LX_199801.pdf from my previous post.
Yes, I know why AGP was introduced, I'm not GenZ, I was also there.

The reason I bought the PCI version of the card was because I was initially planning on using the upcoming version of ITX LLama which will have 1 PCI slot. But it got delayed, and I ended up buying 6ALX instead as a thanksgiving weekend project.

I have looked into that datasheet too and i did not even found the word "lane". 😀

I know the problem with you buying the PCI version because I had a Mini-ITX system too in early 2000's. It was a SV25 / FV25 from Shuttle.
It had a single PCI slot where I used a TV card in. And the chipset has bridged it to its internal AGP bus with its internal chipset graphics.

btw. I'm early GenY.

MattRocks, I did disable parallel and serial ports from the BIOS.
Unfortunately, disabling the FireWire device in the manager did not solve the problem in each case.
I will try the FireWire patches, thanks a lot for digging them up.

I wouldn't rule out a hardware fault on the Audigy card, the problem might be localized to FireWire circuitry with audio still present.

Disruptor wrote on 2025-12-02, 17:24:

I have looked into that datasheet too and i did not even found the word "lane". 😀

Yes, hence the quote-and-quote.

So far, I haven't seen any evidence that sound issues are caused by a bogged down PCI bus, especially given the fact that they continue happening at 640x480 16 color.

hydraphone wrote on 2025-12-02, 15:41:

It is not clear to me whether all PCI slots share the single 32-bit 33MHz interface or not, but if they do, that makes it clear that I need an AGP card for this build.

The 32bit PCI slots:

• share one bus and 100MB/s-133MB/s bandwidth, managed by the chipset.
• are assigned different IRQs, by the PnP OS or BIOS (maybe influenced by chipset).

Your combined cards will be using between 2MB/s (likely) and 40MB/s (max load) of the PCI bandwidth, so the first bullet point is a non-issue.

When I have a system that appears stable in DOS but has various fleeting I/O issues in Windows 98 that come and go, it's often a RAM issue. Run https://www.memtest.org/ overnight. Sometimes memtest+ doesn't find the errors until it runs for 12+ hours.

MattRocks, thanks a lot for this information!
That could be limiting down the road if I add a PCI SATA card into the mixture, and on top the potential IRQ issues, I think it will still be useful for me to get an AGP card to rule out potential hardware issues (PCI slots themselves/Audigy card).

Thanks douglar, I have also been thinking about that as well because the RAM sticks I have are also used. I need to figure out how to run memtest (since I have no FDD or CD-ROM), I will probably use the 86Box .img trick again.

hydraphone wrote on 2025-12-02, 17:29:

MattRocks, I did disable parallel and serial ports from the BIOS. Unfortunately, disabling the FireWire device in the manager di […]
Show full quote

MattRocks, I did disable parallel and serial ports from the BIOS.
Unfortunately, disabling the FireWire device in the manager did not solve the problem in each case.
I will try the FireWire patches, thanks a lot for digging them up.

I wouldn't rule out a hardware fault on the Audigy card, the problem might be localized to FireWire circuitry with audio still present.

https://forums.anandtech.com/threads/acpi-vs- … tandard.285122/

The above adds weight to my recollections that Creative made crackly sound cards. My sincere honest view is that a little crackle is a positive sign because it proves the hardware DSP is activated (and operating within historically observed and reported parameters - even if some don't believe it). If you start from the positive crackly baseline, then give your DSP isolated IRQ and reduce its driver burdens to find a sweet spot with clearer audio then I'd say you win!

There is more to disable in the manager such as Creative SB16 Emulation - it's just layers of complexity and increases odds of IRQ conflicts. You may need to plod through each device in the manager to find IRQs, but you want Audigy DSP to have its own dedicated IRQ. Apparently PCI SLOT#3 is isolated on LX chipset, and you might find IRQ settings for PCI slots in the BIOS.

hydraphone wrote on 2025-12-02, 18:50:

MattRocks, thanks a lot for this information!
That could be limiting down the road if I add a PCI SATA card into the mixture, and on top the potential IRQ issues, I think it will still be useful for me to get an AGP card to rule out potential hardware issues (PCI slots themselves/Audigy card).

Thanks douglar, I have also been thinking about that as well because the RAM sticks I have are also used. I need to figure out how to run memtest (since I have no FDD or CD-ROM), I will probably use the 86Box .img trick again.

The IDE bus has its own dedicated 66-133MB/s bandwidth (depending on chipset) and I don't want to waste that. I did read warnings about bad IDE-SATA adapters, but I don't think that's universally true so I'm going to try IDE-SATA. Nothing to report yet because I'm still watching and lining my ducks up 😉

MattRocks, I am actually more bothered by Windows randomly replaying previously played sounds, like the startup sound or menu right-click sounds or any previously played music.
I'm also looking forward to testing disk throughput once I get sound working normally and get an AGP card!

Some more info from Linux

1cat /proc/interrupts
2
3           CPU0       
4  0:     316285    XT-PIC      timer
5  1:       2358    XT-PIC      i8042
6  2:          0    XT-PIC      cascade
7  8:          0    XT-PIC      rtc0
8 10:          0    XT-PIC      firewire_ohci
9 11:     200000    XT-PIC      uhci_hcd:usb1, snd_emu10k1
10 12:        167    XT-PIC      i8042
11 14:       5441    XT-PIC      ata_piix
12 15:          0    XT-PIC      ata_piix
13NMI:          0   Non-maskable interrupts
14LOC:          0   Local timer interrupts
15SPU:          0   Spurious interrupts
16PMI:          0   Performance monitoring interrupts
17IWI:          0   IRQ work interrupts
18RTR:          0   APIC ICR read retries
19RES:          0   Rescheduling interrupts
20CAL:          0   Function call interrupts
21TLB:          0   TLB shootdowns
22TRM:          0   Thermal event interrupts
23THR:          0   Threshold APIC interrupts
24DFR:          0   Deferred Error APIC interrupts
25MCE:          0   Machine check exceptions
26MCP:          3   Machine check polls
27ERR:        709
28MIS:          0
29PIN:          0   Posted-interrupt notification event
30NPI:          0   Nested posted-interrupt event
31PIW:          0   Posted-interrupt wakeup event

some highlights from dmesg

1[   83.875060] PCI: setting IRQ 10 as level-triggered
2[   83.875110] firewire_ohci 0000:00:0a.2: found PCI INT B -> IRQ 10
3[   83.875110] firewire_ohci 0000:00:0a.2: sharing IRQ 10 with 0000:00:08.0
4[   83.932853] firewire_ohci 0000:00:0a.2: added OHCI v1.10 device as card 0, 4 IR + 8 IT contexts, quirks 0x2
5[   84.652560] irq 11: nobody cared (try booting with the "irqpoll" option)
6[   84.652710] CPU: 0 PID: 9 Comm: kworker/0:1 Not tainted 6.9.7-arch1-1.2 #1 00f4d872e619c93a2fd3dfaa7198b1045c00f8c9
7[   84.652753] Workqueue:  0x0 (firewire)
8[   84.652805] Call Trace:
9[   84.652825]  <IRQ>
10[   84.652846]  dump_stack_lvl+0x41/0x60
11[   84.652909]  dump_stack+0xd/0x12
12[   84.652939]  __report_bad_irq+0x38/0xa4
13[   84.652996]  note_interrupt.cold+0xc/0x57
14[   84.653032]  handle_irq_event+0x69/0x70
15[   84.653035]  ? __irq_disable+0xa0/0xa0
16[   84.653035]  handle_level_irq+0x7a/0x160
17[   84.653035]  __handle_irq+0x52/0x90
18[   84.653035]  </IRQ>
19[   84.653035]  <SOFTIRQ>
20[   84.653035]  __common_interrupt+0x36/0xb0
21[   84.653035]  common_interrupt+0x2a/0x50
22[   84.653035]  asm_common_interrupt+0x102/0x140
23[   84.653035] EIP: handle_softirqs+0x6e/0x250
24[   84.653035] Code: 64 a1 00 2a 3e d0 ba 00 01 00 00 e8 dc 3c 05 00 c7 45 e0 0a 00 00 00 89 75 e4 89 5d dc 31 c0 64 66 a3 14 2a 3e d0 fb 0f bc cf <75> 05 b9 ff ff ff ff 89 cb b8 80 70 15 d0 83 c3 01 74 78 89 c6 89
25[   84.653035] EAX: 00000000 EBX: 04208160 ECX: 00000001 EDX: 00000000
26[   84.653035] ESI: fffcb1e9 EDI: 00000002 EBP: c1113ff0 ESP: c1113fcc
27[   84.653035] DS: 007b ES: 007b FS: 00d8 GS: 0000 SS: 0068 EFLAGS: 00000206
28[   84.653035]  ? handle_softirqs+0x6e/0x250
29[   84.653035]  ? __lock_text_end+0x3/0x3
30[   84.653035]  __do_softirq+0xf/0x13
31[   84.653035]  do_softirq_own_stack+0x26/0x70
32[   84.653035]  </SOFTIRQ>
33[   84.653035]  irq_exit_rcu+0x72/0xb0
34[   84.653035]  common_interrupt+0x2f/0x50
35[   84.653035]  asm_common_interrupt+0x102/0x140
36[   84.653035] EIP: preempt_schedule_irq+0x29/0x60
37[   84.653035] Code: 00 3e 8d 74 26 00 64 f7 05 04 2a 3e d0 ff ff ff 7f 75 3e 9c 58 f6 c4 02 75 37 55 89 e5 64 ff 05 04 2a 3e d0 fb b8 01 00 00 00 <e8> 32 e5 ff ff fa 64 a1 00 2a 3e d0 64 ff 0d 04 2a 3e d0 8b 00 a8
38[   84.653035] EAX: 00000001 EBX: c116def0 ECX: 00000003 EDX: 00000000
39[   84.653035] ESI: 00000030 EDI: 00000000 EBP: c116dec4 ESP: c116dec4
40[   84.653035] DS: 007b ES: 007b FS: 00d8 GS: 0000 SS: 0068 EFLAGS: 00000202
41[   84.653035]  ? preempt_schedule_irq+0x29/0x60
42[   84.653035]  raw_irqentry_exit_cond_resched+0x30/0x40
43[   84.653035]  irqentry_exit+0x1d/0x50
44[   84.653035]  common_interrupt+0x38/0x50
45[   84.653035]  asm_common_interrupt+0x102/0x140
46[   84.653035] EIP: _raw_spin_unlock_irq+0x16/0x30
47[   84.653035] Code: c3 8d 76 00 e8 33 12 00 00 5d 31 c0 31 d2 c3 2e 8d 74 26 00 3e 8d 74 26 00 55 89 e5 c6 00 00 8d 76 00 fb 64 ff 0d 04 2a 3e d0 <74> 08 5d 31 c0 31 d2 c3 66 90 e8 03 12 00 00 5d 31 c0 31 d2 c3 cc
48[   84.653035] EAX: dc9a9bc0 EBX: c110f200 ECX: 00000000 EDX: 00000000
49[   84.653035] ESI: dc9a9bc0 EDI: c110f228 EBP: c116df4c ESP: c116df4c
50[   84.653035] DS: 007b ES: 007b FS: 00d8 GS: 0000 SS: 0068 EFLAGS: 00000286
51[   84.653035]  ? _raw_spin_unlock_irq+0x16/0x30
52[   84.653035]  worker_thread+0x181/0x350
53[   84.653035]  kthread+0xdb/0x110
54[   84.653035]  ? rescuer_thread+0x3d0/0x3d0
55[   84.653035]  ? kthread_complete_and_exit+0x20/0x20
56[   84.653035]  ret_from_fork+0x38/0x60
57[   84.653035]  ? kthread_complete_and_exit+0x20/0x20
58[   84.653035]  ret_from_fork_asm+0x12/0x20
59[   84.653035]  entry_INT80_32+0xf0/0xf0
60[   84.653035] handlers:

61[   84.653035] [<9bffe353>] usb_hcd_irq
62[   84.653035] Disabling IRQ #11
63
64
65
66[  134.756250] nvidiafb 0000:00:08.0: found PCI INT A -> IRQ 10
67[  134.760295] nvidiafb 0000:00:08.0: sharing IRQ 10 with 0000:00:0a.2
68[  134.769064] nvidiafb: Device ID: 10de0322 
69[  134.770610] nvidiafb 0000:00:08.0: vgaarb: deactivate vga console
70[  134.775386] Console: switching to colour dummy device 80x25
71[  134.799241] nvidiafb: CRTC0 analog found
72[  134.806482] nvidiafb: CRTC1 analog not found
73[  135.363385] nvidiafb: EDID found from BUS1
74[  135.471423] i2c i2c-2: unable to read EDID block.
75[  135.628349] i2c i2c-2: unable to read EDID block.
76[  135.799098] i2c i2c-2: unable to read EDID block.
77[  135.849555] nvidiafb: CRTC 0 appears to have a CRT attached
78[  135.849616] nvidiafb: Using CRT on CRTC 0
79[  135.868065] fbcon: NV32 (fb0) is primary device
80[  136.060117] Console: switching to colour frame buffer device 240x67
81[  136.095408] nvidiafb: PCI nVidia NV32 framebuffer (64MB @ 0xD0000000)
82[  139.298974] snd_emu10k1 0000:00:0a.0: found PCI INT A -> IRQ 11
83[  139.299076] snd_emu10k1 0000:00:0a.0: sharing IRQ 11 with 0000:00:07.2
84[  140.099050] irq 11: nobody cared (try booting with the "irqpoll" option)
85[  140.099050] CPU: 0 PID: 265 Comm: (udev-worker) Not tainted 6.9.7-arch1-1.2 #1 00f4d872e619c93a2fd3dfaa7198b1045c00f8c9
86[  140.099050] Call Trace:
87[  140.099050]  <IRQ>
88[  140.099050]  dump_stack_lvl+0x41/0x60
89[  140.099050]  dump_stack+0xd/0x12
90[  140.099050]  __report_bad_irq+0x38/0xa4
91[  140.099050]  note_interrupt.cold+0xc/0x57
92[  140.099050]  handle_irq_event+0x69/0x70
93[  140.099050]  ? __irq_disable+0xa0/0xa0
94[  140.099050]  handle_level_irq+0x7a/0x160
95[  140.099050]  __handle_irq+0x52/0x90
96[  140.099050]  </IRQ>
97[  140.099050]  <SOFTIRQ>
98[  140.099050]  __common_interrupt+0x36/0xb0
99[  140.099050]  common_interrupt+0x2a/0x50
100[  140.099050]  asm_common_interrupt+0x102/0x140
101[  140.099050] EIP: handle_softirqs+0x6e/0x250
102[  140.099050] Code: 64 a1 00 2a 3e d0 ba 00 01 00 00 e8 dc 3c 05 00 c7 45 e0 0a 00 00 00 89 75 e4 89 5d dc 31 c0 64 66 a3 14 2a 3e d0 fb 0f bc cf <75> 05 b9 ff ff ff ff 89 cb b8 80 70 15 d0 83 c3 01 74 78 89 c6 89
103[  140.099050] EAX: 00000000 EBX: 00400140 ECX: 00000001 EDX: 00000000
104[  140.099050] ESI: fffd8a3a EDI: 00000002 EBP: c1113ff0 ESP: c1113fcc
105[  140.099050] DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068 EFLAGS: 00000206
106[  140.099050]  ? __irq_alloc_descs+0x2db/0x320
107[  140.099050]  ? handle_softirqs+0x6e/0x250
108[  140.099050]  ? __lock_text_end+0x3/0x3
109[  140.099050]  __do_softirq+0xf/0x13
110[  140.099050]  do_softirq_own_stack+0x26/0x70
111[  140.099050]  </SOFTIRQ>
112[  140.099050]  irq_exit_rcu+0x72/0xb0
113[  140.099050]  common_interrupt+0x2f/0x50
114[  140.099050]  asm_common_interrupt+0x102/0x140
115[  140.099050] EIP: preempt_schedule_irq+0x29/0x60
116[  140.099050] Code: 00 3e 8d 74 26 00 64 f7 05 04 2a 3e d0 ff ff ff 7f 75 3e 9c 58 f6 c4 02 75 37 55 89 e5 64 ff 05 04 2a 3e d0 fb b8 01 00 00 00 <e8> 32 e5 ff ff fa 64 a1 00 2a 3e d0 64 ff 0d 04 2a 3e d0 8b 00 a8
117[  140.099050] EAX: 00000001 EBX: c2d7fad8 ECX: 00000003 EDX: 00000000
118[  140.099050] ESI: 00000030 EDI: 00000000 EBP: c2d7faac ESP: c2d7faac
119[  140.099050] DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068 EFLAGS: 00000202
120[  140.099050]  ? preempt_schedule_irq+0x29/0x60
121[  140.099050]  raw_irqentry_exit_cond_resched+0x30/0x40
122[  140.099050]  irqentry_exit+0x1d/0x50
123[  140.099050]  common_interrupt+0x38/0x50
124[  140.099050]  asm_common_interrupt+0x102/0x140
125[  140.099050] EIP: _raw_spin_unlock_irqrestore+0x1d/0x40
126[  140.099050] Code: e8 b0 18 00 00 31 c0 5d 31 d2 31 c9 c3 3e 8d 74 26 00 55 89 d1 89 e5 c6 00 00 8d 76 00 80 e5 02 74 01 fb 64 ff 0d 04 2a 3e d0 <74> 11 5d 31 c0 31 d2 31 c9 c3 2e 8d b4 26 00 00 00 00 90 e8 73 18
127[  140.099050] EAX: c110ee70 EBX: c110ee00 ECX: 00000282 EDX: 00000282
128[  140.099050] ESI: c3ecd5c0 EDI: c3e05040 EBP: c2d7fb34 ESP: c2d7fb34
129[  140.099050] DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068 EFLAGS: 00000286
130[  140.099050]  ? _raw_spin_unlock_irqrestore+0x1d/0x40
131[  140.099050]  __setup_irq+0x2f5/0x5f0
132[  140.099050]  request_threaded_irq+0xc1/0x190
133[  140.099050]  devm_request_threaded_irq+0x54/0xb0
134[  140.099050]  ? snd_emu10k1_resume_regs+0xe0/0xe0 [snd_emu10k1 3249cfd8d98bb8ec94b1471be92d0cf61aa675c9]
135[  140.099050]  snd_emu10k1_create+0x429/0x720 [snd_emu10k1 3249cfd8d98bb8ec94b1471be92d0cf61aa675c9]
136[  140.099050]  snd_card_emu10k1_probe+0xd1/0x3c0 [snd_emu10k1 3249cfd8d98bb8ec94b1471be92d0cf61aa675c9]
137[  140.099050]  pci_device_probe+0xa3/0x190
138[  140.099050]  really_probe+0xc2/0x310
139[  140.099050]  ? pm_runtime_barrier+0x51/0x90
140[  140.099050]  __driver_probe_device+0x71/0x140
141[  140.099050]  driver_probe_device+0x1f/0xb0
142[  140.099050]  __driver_attach+0xb7/0x1c0
143[  140.099050]  ? __device_attach_driver+0x110/0x110
144[  140.099050]  bus_for_each_dev+0x62/0xb0
145[  140.099050]  driver_attach+0x19/0x20
146[  140.099050]  ? __device_attach_driver+0x110/0x110
147[  140.099050]  bus_add_driver+0xe6/0x1c0
148[  140.099050]  ? pci_bus_num_vf+0x20/0x20
149[  140.099050]  driver_register+0x6d/0xd0
150[  140.099050]  __pci_register_driver+0x41/0x50
151[  140.099050]  ? 0xe0dd2000
152[  140.099050]  emu10k1_driver_init+0x1c/0x1000 [snd_emu10k1 3249cfd8d98bb8ec94b1471be92d0cf61aa675c9]
153[  140.099050]  do_one_initcall+0x40/0x2b0
154[  140.099050]  ? kfence_guarded_free+0x125/0x200
155[  140.099050]  ? kmalloc_trace+0xfe/0x2f0
156[  140.099050]  ? do_init_module+0x21/0x1f0
157[  140.099050]  do_init_module+0x55/0x1f0
158[  140.099050]  load_module+0x17da/0x19e0
159[  140.099050]  init_module_from_file+0x71/0xb0
160[  140.099050]  __ia32_sys_finit_module+0x11b/0x280
161[  140.099050]  ia32_sys_call+0x1b63/0x21d0
162[  140.099050]  do_int80_syscall_32+0x52/0x90
163[  140.099050]  ? get_stack_info+0x140/0x140
164[  140.099050]  ? enable_8259A_irq+0x5c/0x70
165[  140.099050]  ? handle_level_irq+0xa2/0x160
166[  140.099050]  ? irqentry_exit_to_user_mode+0x63/0x180
167[  140.099050]  ? irqentry_exit+0x45/0x50
168[  140.099050]  ? common_interrupt+0x38/0x50
169[  140.099050]  entry_INT80_32+0xf0/0xf0
170[  140.099050] EIP: 0xaf4d65d2
171[  140.099050] Code: f0 ff c3 66 90 66 90 90 55 57 56 53 8b 6c 24 2c 8b 7c 24 28 8b 74 24 24 8b 54 24 20 8b 4c 24 1c 8b 5c 24 18 8b 44 24 14 cd 80 <5b> 5e 5f 5d 3d 01 f0 ff ff 0f 83 1f d0 f0 ff c3 66 90 66 90 66 90
172[  140.099050] EAX: ffffffda EBX: 00000022 ECX: af5de3f4 EDX: 00000004
173[  140.099050] ESI: 113bc810 EDI: 113b3c70 EBP: bf9f5f28 ESP: bf9f5ecc
174[  140.099050] DS: 007b ES: 007b FS: 0000 GS: 0033 SS: 007b EFLAGS: 00000292
175[  140.099050] handlers:
176[  140.099050] [<9bffe353>] usb_hcd_irq
177[  140.099050] [<09cb4f32>] snd_emu10k1_interrupt [snd_emu10k1]
178[  140.099050] Disabling IRQ #11
179[  140.109981] snd_emu10k1 0000:00:0a.0: Installing spdif_bug patch: SB Audigy 2 ZS [SB0350]

So far, I haven't been able to reproduce any of the sound issues on Linux.

No X11 yet, so far. nouveau refuses to load, because it depends on wmi, which claims no device, might be related to ACPI issues.

1[    0.058175] ACPI: Early table checksum verification disabled
2[    0.066161] ACPI BIOS Error (bug): A valid RSDP was not found (20230628/tbxfroot-222)

nvidiafb thinks it's connected to a CRT. LCD doesn't like the signal by vesa (with nomodeset) or fbdev with automatically detected settings by X11. Might try to figure out the correct modeline info for it, but it's probably not worth it. Planning to buy an actual CRT soon, should work with it automatically.

MattRocks wrote on 2025-12-02, 11:25:

Using Archive.org I was able to download "something" from the URL that claims to combine all the Win98SE Firewire+USB patches.

There are a few captures to explore: https://web.archive.org/web/20170904074655/ht … om/spx/NUSB.EXE

Attached whatever landed in my Downloads folder. Use at your own risk: I cannot promise this works. I haven't used it myself - and I haven't even tried to extract it so to check this isn't just a HTML broken link error!

Tried it, and re-enabled the FireWire device. Unfortunately, it freezes at Windows splash screen as usual.

hydraphone wrote on 2025-12-02, 20:59:

MattRocks, I am actually more bothered by Windows randomly replaying previously played sounds, like the startup sound or menu right-click sounds or any previously played music.

That screams of IRQ sharing. Win98 assumes ACPI (virtual and shared IRQs) is safe. ACPI solved one problem where devices fight over IRQs (infamous Win95 IRQ conflicts), and ACPI creates the problem of denying devices an isolated IRQ when they need one. Your Audigy definitely needs its own IRQ.

Here's the cleanest ACPI next step I found, with a foretelling of what to expect immediately afterwards..
Re: Windows 98se reboots on shutdown

That sounds promising, I gave it a try, but it didn't seem to have any effect. I tried rebooting, but none of the devices were removed. I manually removed the Audigy components + USB + NVIDIA, followed by reboots, and the devices got added with their old configuration. The PCI to USB Universal Host Controller and the Creative Audigy Audio Processor still both use IRQ 11 for example.

I also tried to add ACPIOption as string but no luck. Did I miss doing something?

I noticed that in the device manager, under System devices, I see "Advanced Power Management Support" and no mention of ACPI, so it appears I never had ACPI on in the first place?
I don't see it in the BIOS or the motherboard's manual either, so it appears this is an APM only PC.

Put Audigy in PCI slot #1 (and skipping over FireWire detection), and NVDIA in #2: sound works flawlessly, NVIDIA is again stuck at 640x480 at 16 colors, any changes results in reboot/locks...
Adding FireWire results in immediate reboot/lockup.

Minor progress.

After an hour of mindless fiddling around with the BIOS options and PCI slot arrangements (at some point, I manually disabled some IRQs and manually assigned IRQs to slots from BIOS, and then reverted the settings), both sound and video (1280x720@32bpp) worked fine... until Windows froze after playing music for about a minute. Reboot and same thing happening again.

NVIDIA at PCI slot #1, Audigy at #4. Windows assigned IRQ #10 to both NVIDIA and Audigy, and USB Universal Host Controller got #11.

Switched the resolution to 1024x768@16bpp, still locking up in a minute or so.

Tried disabling the USB Universal Host Controller from the device manager, 5 minutes of music and no lock ups so far. I lost the USB inputs now, but well, that's so far the longest I enjoyed this PC!

hydraphone wrote on 2025-12-03, 04:57:

that's so far the longest I enjoyed this PC!

Congratulations!
Before starting a long gaming session (especially in games without auto-save), I would check the stability.

It seems that 3DMark 2003 does not require SSE instructions.
Run the looped test for an hour with GPU+CPU+60 sounds.