Glad to hear that. It could be a failing chipset on the other board then. Would be interesting to test the Geforce 4 MX from your other thread on this board to see if it still has memory errors.

asdf53 wrote on 2025-10-14, 15:21:

Glad to hear that. It could be a failing chipset on the other board then. Would be interesting to test the Geforce 4 MX from your other thread on this board to see if it still has memory errors.

That's not a bad idea. Apart from that though is there anything else I could test?

Just tested the Geforce 4 with it - same issue. Has broken graphics/lines. Definitely knackered, that.

DustyShinigami wrote on 2025-10-14, 15:29:

That's not a bad idea. Apart from that though is there anything else I could test?

Here are a couple of ideas:

Prime95 to test the CPU and cache (prime95 v21 for win9x), go to Options -> Torture Test

MemTest to test RAM in Windows

Run 3DMark in loop mode for 15-30 minutes to test GPU and Northbridge

For the Southbridge (handles the PCI stuff), can't think of any ready made tools - basically anything that involves lots of data transfer between PCI devices. Maybe enabling DMA mode and copying large files around, for example from USB to hard disk and vice versa, or copy a large zip file from a network drive onto the hard disk, unzip it and see if it complains about errors. That's a typical thing that would happen.

asdf53 wrote on 2025-10-14, 17:15:

Here are a couple of ideas: […]
Show full quote

DustyShinigami wrote on 2025-10-14, 15:29:

That's not a bad idea. Apart from that though is there anything else I could test?

Here are a couple of ideas:

Prime95 to test the CPU and cache (prime95 v21 for win9x), go to Options -> Torture Test

MemTest to test RAM in Windows

Run 3DMark in loop mode for 15-30 minutes to test GPU and Northbridge

For the Southbridge (handles the PCI stuff), can't think of any ready made tools - basically anything that involves lots of data transfer between PCI devices. Maybe enabling DMA mode and copying large files around, for example from USB to hard disk and vice versa, or copy a large zip file from a network drive onto the hard disk, unzip it and see if it complains about errors. That's a typical thing that would happen.

Okay. How long should I let Prime95 run for? I'm not sure what to say about a network drive. Haven't set one up and used one before. I'm always sending files from a USB pen drive though. The last biggest file I copied over was an image of the Sound Blaster's driver CD. It's not given me any issues so far.

Also, so far, it's recognised and installed the drivers for the Geforce 4 Ti 4200. So running 3DMark is definitely a plan. 😀

DustyShinigami wrote on 2025-10-14, 17:47:

Okay. How long should I let Prime95 run for? I'm not sure what to say about a network drive. Haven't set one up and used one before. I'm always sending files from a USB pen drive though. The last biggest file I copied over was an image of the Sound Blaster's driver CD. It's not given me any issues so far.

Also, so far, it's recognised and installed the drivers for the Geforce 4 Ti 4200. So running 3DMark is definitely a plan. 😀

If no errors after 20 minutes, it should be good. You don't need a network drive, it was just an example of something that involves PCI and is sensitive to corruption. USB should also be good, or even copying the file from one folder to another on the same hard disk. It only needs to be large enough to make errors more likely, several hundred MB would be good. Verification after copying can be done with a tool such as HashMyFiles, and as mentioned, make sure that DMA mode is enabled.

asdf53 wrote on 2025-10-14, 18:12:

DustyShinigami wrote on 2025-10-14, 17:47:

Okay. How long should I let Prime95 run for? I'm not sure what to say about a network drive. Haven't set one up and used one before. I'm always sending files from a USB pen drive though. The last biggest file I copied over was an image of the Sound Blaster's driver CD. It's not given me any issues so far.

Also, so far, it's recognised and installed the drivers for the Geforce 4 Ti 4200. So running 3DMark is definitely a plan. 😀

If no errors after 20 minutes, it should be good. You don't need a network drive, it was just an example of something that involves PCI and is sensitive to corruption. USB should also be good, or even copying the file from one folder to another on the same hard disk. It only needs to be large enough to make errors more likely, several hundred MB would be good. Verification after copying can be done with a tool such as HashMyFiles, and as mentioned, make sure that DMA mode is enabled.

Awesome. Thank you.

It's currently running through 3DMark 2000 at the moment.

Okay, it successfully completed the default 3DMark test, however when I set it up to use higher settings and on a loop, I think it crashed. Not sure if because of the GPU or if it was just a simple application crash. I left it to go through the Video Memory Test under Windows.

Well, this is setting off my paranoia. First the crash with 3DMark and now when I came back to see if VMST had finished, the system had hung up. This is something that's happened before - when I come back and try to wake the machine from standby and it doesn't. The monitor light just flashes. It doesn't respond to keyboard or mouse click presses. I think someone did suggest something to disable in the BIOS, but I can't remember what. So I'm running through VMST again.

Second attempt and it did it again. PC/monitor was in standby and wouldn't come out of it. Had to force it to shutdown. One other thing I've noticed is that it no longer tries to go through ScanDisk. And yet the option to disable it hasn't been enabled.

Going through it a third time but this time with Power options, like Standby etc, disabled and the monitor left on.

No one promised stable operation of the 440BX at 133 MHz FSB.
Moreover, the 4MX is not required to run at 89 MHz AGP, although it most often does.
Try 100 MHz FSB.
Or 124 MHz (PСI - 31 MHz).

shevalier wrote on 2025-10-15, 04:29:

No one promised stable operation of the 440BX at 133 MHz FSB. Moreover, the 4MX is not required to run at 89 MHz AGP, although i […]
Show full quote

No one promised stable operation of the 440BX at 133 MHz FSB.
Moreover, the 4MX is not required to run at 89 MHz AGP, although it most often does.
Try 100 MHz FSB.
Or 124 MHz (PСI - 31 MHz).

I believe this one does though, at least according to the seller I bought it off. They were running the CPU at 933MHz, which is what I’ve set it too.

I’ve completed stress testing the GPU anyway and everything was fine. I also overclocked it to better match the default clock speeds of the card (the make of this card has it slightly under clocked) Tests were are fine. I’ll share the results a bit later.

Right, so these are the results from VMT. The two errors were because of it going into standby and me not being able to get out of it. Not sure what triggers that. But the final test was fine.

Changing video mode to 640x480x16...OK WARNING: Do not place other windows over program window! INFORMATION: Only one mode can b […]
Show full quote

Changing video mode to 640x480x16...OK
WARNING: Do not place other windows over program window!
INFORMATION: Only one mode can be tested now, all others will not match.
[14/10/25 07:34:00] Test started for "Primary Display Driver (NVIDIA GeForce4 Ti 4200 with AGP8X)"...
Trying 16bpp RGB:565 mode...OK
TEST FAIL (Code: 887601AE)
Trying 16bpp RGB:555 mode...NOT MATCH WITH CURRENT
Trying 16bpp BGR:565 mode...NOT MATCH WITH CURRENT
Trying 32bpp RGB:888 mode...NOT MATCH WITH CURRENT
Trying 32bpp BGR:888 mode...NOT MATCH WITH CURRENT
NO PASS COMPLETED
Changing video mode to 640x480x16...OK
[14/10/25 09:10:49] Test started for "Primary Display Driver (NVIDIA GeForce4 Ti 4200 with AGP8X)"...
Trying 16bpp RGB:565 mode...OK
Trying 16bpp RGB:555 mode...OK
TEST FAIL (Code: 887601AE)
Trying 16bpp BGR:565 mode...OK
TEST FAIL (Code: 887601AE)
Trying 32bpp RGB:888 mode...OK
TEST FAIL (Code: 887601AE)
Trying 32bpp BGR:888 mode...OK
TEST FAIL (Code: 887601AE)
[14/10/25 09:25:54] Pass completed (0 errors found).
Changing video mode to 640x480x16...OK
WARNING: Do not place other windows over program window!
INFORMATION: Only one mode can be tested now, all others will not match.
[14/10/25 10:47:48] Test started for "Primary Display Driver (NVIDIA GeForce4 Ti 4200 with AGP8X)"...
Trying 16bpp RGB:565 mode...OK
Trying 16bpp RGB:555 mode...NOT MATCH WITH CURRENT
Trying 16bpp BGR:565 mode...NOT MATCH WITH CURRENT
Trying 32bpp RGB:888 mode...NOT MATCH WITH CURRENT
Trying 32bpp BGR:888 mode...NOT MATCH WITH CURRENT
[14/10/25 11:10:22] Pass completed (0 errors found).

And then the default result from 3DMark 2000:

3DMark Database File Name: Database1.3DB Database Name: Database1 Number of Projects in Database: 1 PROJECT Description: My […]
Show full quote

3DMark Database
File Name: Database1.3DB
Database Name: Database1
Number of Projects in Database: 1

PROJECT
Description: My Project
Comment:
3DMark Version: 340
Install ID: 0xCD629384
Registration Key: 6VZTY-SDZ75-KA9V7
Registration Name: Futuremark
Registered: Yes

TEMPLATE
Display:
Type: Internal
Width: 1024
Height: 768
Depth: 16-Bit
Buffering: Triple
Z-Buffering: 16-Bit
Refresh Rate: VSync Off
Texture Format: 16-Bit

Options:
Continuous Benchmark: No
Show Title Screens: Yes
Benchmark Run Count: 1
Continuous Demo Mode: No
Sounds Enabled: Yes
CPU Optimization: D3D Hardware T&L

RESULTS
Platform: Internal
3DMark Result: 5775 3D marks
CPU Speed: 407 CPU 3D marks
Game 1 - Helicopter - Low Detail: 104.8 FPS
Game 1 - Helicopter - Medium Detail: 78.6 FPS
Game 1 - Helicopter - High Detail: 57.1 FPS
Game 2 - Adventure - Low Detail: 143.9 FPS
Game 2 - Adventure - Medium Detail: 60.7 FPS
Game 2 - Adventure - High Detail: 36.3 FPS
Fill Rate (Single-Texturing): 883.9 MTexels/s
Fill Rate (Multi-Texturing): 1778.3 MTexels/s
High Polygon Count (1 Light): 14528 KTriangles/s
High Polygon Count (4 Lights): 8570 KTriangles/s
High Polygon Count (8 Lights): 4532 KTriangles/s
8MB Texture Rendering Speed: 519.6 FPS
16MB Texture Rendering Speed: 274.5 FPS
32MB Texture Rendering Speed: 150.9 FPS
64MB Texture Rendering Speed: 76.4 FPS
Bump Mapping (Emboss, 3-pass): 232.6 FPS
Bump Mapping (Emboss, 2-pass): 339.3 FPS
Bump Mapping (Emboss, 1-pass): 631.6 FPS
Bump Mapping (Environment): 549.2 FPS

SYSTEM
Software Versions:
Windows Version Windows 4 A , Build 2222
DirectX Version 4.07.00.0700
Bios Version Award Modular BIOS v6.00PG
Bios Date 12/03/01

Memory:
Total Physical Memory 256MB
Free Physical Memory 71MB

PROCESSOR
CPU Information:
Processor Type Intel Pentium III
Processor Speed 950MHz
Processor Caps SSE, MMX

Cache Memory:
L1 Cache Size 32KB
L2 Cache Size 256KB

DESKTOP
2D Display Adapter:
Name NVIDIA GeForce4 Ti 4200 with AGP8X
Driver Date 7-28-2003

Monitor:
Name Default Monitor
Driver Date 4-23-1999

Desktop Size and Color Depth:
Width 800
Height 600
Color Depth 16 bit

INTERNAL 3D CARD
Information:
Name NVIDIA GeForce4 Ti 4200 with AGP8X
Driver Name NVDD32.DLL
Driver Version 4.14.10.4523 (Certified)
Total Video Memory 130298KB
Total Texture Memory 162128KB
Minimum Texture Resolution 1 * 1
Maximum Texture Resolution 4096 * 4096
Maximum Texture Blending Stages 8
Maximum Texture Coordinates 8
Bus AGP
Manufacturer Identifier 0x000010DE
Chipset Type 0x00000281
Subsystem Identifier 0x88901462
Chipset Revision Level 0x000000A1

Supported features:
Transformation and lighting in hardware Yes
16-Bit Rendering Yes
24-Bit Rendering No
32-Bit Rendering Yes
Textures In Single Pass 4
Driver can disable VSync by software request Yes
Edge Antialiasing Yes
Subpixel Accuracy Yes
Point Sampling Yes
Point Sampling With Mip-Mapping Yes
Bilinear Filtering Yes
Bilinear Filtering With Mip-Mapping Yes
Trilinear Filtering Yes
Anisotropic Filtering No
Specular Gouraud Yes
Vertex Fog Yes
Range Fog Yes
Table Fog Yes
W-Fog Yes
Alpha Blending Yes
Vertex Alpha Blending Yes
Vertex And Texture Alpha Blending Yes
Additive Alpha Blending Yes
Multiplicative Alpha Blending Yes
S3 Texture Compression Yes
DX6 Bump Mapping Yes

Flags:
DirectDraw flags 1 0x94C27FE3
DirectDraw flags 2 0x206AFE7D
Texture operations 0x03FEFFFF
Stencil-buffer operations 0x000000FF
Miscellaneous capabilities 0x0000AEF2
Raster capabilities 0x07F77DB1
Z-comparison capabilities 0x000000FF
Alpha-test-comparison capabilities 0x000000FF
Source-blending capabilities 0x00003FFF
Destination-blending capabilities 0x00003FFF
Shading capabilities 0x000C528F
Texture capabilities 0x0007ED8F
Texture-filtering capabilities 0x0303073F
Texture-blending capabilities 0x000000FF
Texture-addressing capabilities 0x0000001F

Modes:
Z-Buffer formats 16-Bit Z-Buffer, 24-Bit Z-Buffer, 32-Bit Z-Buffer (8-Bit Stencil)
8-Bit Texture formats N/A
15-Bit Texture formats 15-bit, 555 RGB
16-Bit Texture formats 16-bit, 5551 RGBA, 16-bit, 4444 RGBA, 16-bit, 565 RGB
24-Bit Texture formats 24-bit, 888 RGB
32-Bit Texture formats 32-bit, 8888 RGBA
Display modes 320 * 200, 8 bit, 320 * 240, 8 bit, 400 * 300, 8 bit, 480 * 360, 8 bit, 512 * 384, 8 bit, 640 * 400, 8 bit, 640 * 480, 8 bit, 720 * 480, 8 bit, 720 * 576, 8 bit, 800 * 600, 8 bit, 848 * 480, 8 bit, 960 * 720, 8 bit, 1024 * 768, 8 bit, 1152 * 864, 8 bit, 1280 * 720, 8 bit, 1280 * 768, 8 bit, 1280 * 960, 8 bit, 1280 * 1024, 8 bit, 320 * 200, 16 bit, 320 * 240, 16 bit, 400 * 300, 16 bit, 480 * 360, 16 bit, 512 * 384, 16 bit, 640 * 400, 16 bit, 640 * 480, 16 bit, 720 * 480, 16 bit, 720 * 576, 16 bit, 800 * 600, 16 bit, 848 * 480, 16 bit, 960 * 720, 16 bit, 1024 * 768, 16 bit, 1152 * 864, 16 bit, 1280 * 720, 16 bit, 1280 * 768, 16 bit, 1280 * 960, 16 bit, 1280 * 1024, 16 bit, 320 * 200, 32 bit, 320 * 240, 32 bit, 400 * 300, 32 bit, 480 * 360, 32 bit, 512 * 384, 32 bit, 640 * 400, 32 bit, 640 * 480, 32 bit, 720 * 480, 32 bit, 720 * 576, 32 bit, 800 * 600, 32 bit, 848 * 480, 32 bit, 960 * 720, 32 bit, 1024 * 768, 32 bit, 1152 * 864, 32 bit, 1280 * 720, 32 bit, 1280 * 768, 32 bit, 1280 * 960, 32 bit, 1280 * 1024, 32 bit

Please attach description of Your testing environment to all published results, charts and figures. Description is available for all projects. Exporting instructions can be found in Result Browser Help. Publishing requirements are described in the 3DMark license agreement You have agreed on.

The results are created with 3DMark2000(tm) (www.madonion.com). Tests are performed without independent verification by Futuremark Corporation and that Futuremark Corporation makes no representations or warranties as to the result of the test. All products used in the test were shipping versions available to the general public.
3DMark is a trademark of MadOnion.com

And then after it was overclocked to roughly the card's spec:

3DMark Database File Name: Database1.3DB Database Name: Database1 Number of Projects in Database: 1 PROJECT Description: My […]
Show full quote

3DMark Database
File Name: Database1.3DB
Database Name: Database1
Number of Projects in Database: 1

PROJECT
Description: My Project
Comment:
3DMark Version: 340
Install ID: 0xCD629384
Registration Key: 6VZTY-SDZ75-KA9V7
Registration Name: Futuremark
Registered: Yes

TEMPLATE
Display:
Type: Internal
Width: 1024
Height: 768
Depth: 16-Bit
Buffering: Triple
Z-Buffering: 16-Bit
Refresh Rate: VSync Off
Texture Format: 16-Bit

Options:
Continuous Benchmark: No
Show Title Screens: Yes
Benchmark Run Count: 3
Continuous Demo Mode: No
Sounds Enabled: Yes
CPU Optimization: D3D Hardware T&L

RESULTS
Platform: Internal
3DMark Result: 5863 3D marks
CPU Speed: 436 CPU 3D marks
Game 1 - Helicopter - Low Detail: 105.0 FPS
Game 1 - Helicopter - Medium Detail: 79.2 FPS
Game 1 - Helicopter - High Detail: 58.9 FPS
Game 2 - Adventure - Low Detail: 146.6 FPS
Game 2 - Adventure - Medium Detail: 61.8 FPS
Game 2 - Adventure - High Detail: 37.0 FPS
Fill Rate (Single-Texturing): 983.5 MTexels/s
Fill Rate (Multi-Texturing): 1974.4 MTexels/s
High Polygon Count (1 Light): 14683 KTriangles/s
High Polygon Count (4 Lights): 9510 KTriangles/s
High Polygon Count (8 Lights): 5036 KTriangles/s
8MB Texture Rendering Speed: 535.3 FPS
16MB Texture Rendering Speed: 280.1 FPS
32MB Texture Rendering Speed: 151.5 FPS
64MB Texture Rendering Speed: 77.8 FPS
Bump Mapping (Emboss, 3-pass): 237.0 FPS
Bump Mapping (Emboss, 2-pass): 351.6 FPS
Bump Mapping (Emboss, 1-pass): 655.2 FPS
Bump Mapping (Environment): 613.2 FPS

SYSTEM
Software Versions:
Windows Version Windows 4 A , Build 2222
DirectX Version 4.07.00.0700
Bios Version Award Modular BIOS v6.00PG
Bios Date 12/03/01

Memory:
Total Physical Memory 256MB
Free Physical Memory 179MB

PROCESSOR
CPU Information:
Processor Type Intel Pentium III
Processor Speed 950MHz
Processor Caps SSE, MMX

Cache Memory:
L1 Cache Size 32KB
L2 Cache Size 256KB

DESKTOP
2D Display Adapter:
Name NVIDIA GeForce4 Ti 4200 with AGP8X
Driver Date 7-28-2003

Monitor:
Name Default Monitor
Driver Date 4-23-1999

Desktop Size and Color Depth:
Width 1024
Height 768
Color Depth 16 bit

INTERNAL 3D CARD
Information:
Name NVIDIA GeForce4 Ti 4200 with AGP8X
Driver Name NVDD32.DLL
Driver Version 4.14.10.4523 (Certified)
Total Video Memory 130432KB
Total Texture Memory 161664KB
Minimum Texture Resolution 1 * 1
Maximum Texture Resolution 4096 * 4096
Maximum Texture Blending Stages 8
Maximum Texture Coordinates 8
Bus AGP
Manufacturer Identifier 0x000010DE
Chipset Type 0x00000281
Subsystem Identifier 0x88901462
Chipset Revision Level 0x000000A1

Supported features:
Transformation and lighting in hardware Yes
16-Bit Rendering Yes
24-Bit Rendering No
32-Bit Rendering Yes
Textures In Single Pass 4
Driver can disable VSync by software request Yes
Edge Antialiasing Yes
Subpixel Accuracy Yes
Point Sampling Yes
Point Sampling With Mip-Mapping Yes
Bilinear Filtering Yes
Bilinear Filtering With Mip-Mapping Yes
Trilinear Filtering Yes
Anisotropic Filtering No
Specular Gouraud Yes
Vertex Fog Yes
Range Fog Yes
Table Fog Yes
W-Fog Yes
Alpha Blending Yes
Vertex Alpha Blending Yes
Vertex And Texture Alpha Blending Yes
Additive Alpha Blending Yes
Multiplicative Alpha Blending Yes
S3 Texture Compression Yes
DX6 Bump Mapping Yes

Flags:
DirectDraw flags 1 0x94C27FE3
DirectDraw flags 2 0x206AFE7D
Texture operations 0x03FEFFFF
Stencil-buffer operations 0x000000FF
Miscellaneous capabilities 0x0000AEF2
Raster capabilities 0x07F77DB1
Z-comparison capabilities 0x000000FF
Alpha-test-comparison capabilities 0x000000FF
Source-blending capabilities 0x00003FFF
Destination-blending capabilities 0x00003FFF
Shading capabilities 0x000C528F
Texture capabilities 0x0007ED8F
Texture-filtering capabilities 0x0303073F
Texture-blending capabilities 0x000000FF
Texture-addressing capabilities 0x0000001F

Modes:
Z-Buffer formats 16-Bit Z-Buffer, 24-Bit Z-Buffer, 32-Bit Z-Buffer (8-Bit Stencil)
8-Bit Texture formats N/A
15-Bit Texture formats 15-bit, 555 RGB
16-Bit Texture formats 16-bit, 5551 RGBA, 16-bit, 4444 RGBA, 16-bit, 565 RGB
24-Bit Texture formats 24-bit, 888 RGB
32-Bit Texture formats 32-bit, 8888 RGBA
Display modes 320 * 200, 8 bit, 320 * 240, 8 bit, 400 * 300, 8 bit, 480 * 360, 8 bit, 512 * 384, 8 bit, 640 * 400, 8 bit, 640 * 480, 8 bit, 720 * 480, 8 bit, 720 * 576, 8 bit, 800 * 600, 8 bit, 848 * 480, 8 bit, 960 * 720, 8 bit, 1024 * 768, 8 bit, 1152 * 864, 8 bit, 1280 * 720, 8 bit, 1280 * 768, 8 bit, 1280 * 960, 8 bit, 1280 * 1024, 8 bit, 320 * 200, 16 bit, 320 * 240, 16 bit, 400 * 300, 16 bit, 480 * 360, 16 bit, 512 * 384, 16 bit, 640 * 400, 16 bit, 640 * 480, 16 bit, 720 * 480, 16 bit, 720 * 576, 16 bit, 800 * 600, 16 bit, 848 * 480, 16 bit, 960 * 720, 16 bit, 1024 * 768, 16 bit, 1152 * 864, 16 bit, 1280 * 720, 16 bit, 1280 * 768, 16 bit, 1280 * 960, 16 bit, 1280 * 1024, 16 bit, 320 * 200, 32 bit, 320 * 240, 32 bit, 400 * 300, 32 bit, 480 * 360, 32 bit, 512 * 384, 32 bit, 640 * 400, 32 bit, 640 * 480, 32 bit, 720 * 480, 32 bit, 720 * 576, 32 bit, 800 * 600, 32 bit, 848 * 480, 32 bit, 960 * 720, 32 bit, 1024 * 768, 32 bit, 1152 * 864, 32 bit, 1280 * 720, 32 bit, 1280 * 768, 32 bit, 1280 * 960, 32 bit, 1280 * 1024, 32 bit

Please attach description of Your testing environment to all published results, charts and figures. Description is available for all projects. Exporting instructions can be found in Result Browser Help. Publishing requirements are described in the 3DMark license agreement You have agreed on.

The results are created with 3DMark2000(tm) (www.madonion.com). Tests are performed without independent verification by Futuremark Corporation and that Futuremark Corporation makes no representations or warranties as to the result of the test. All products used in the test were shipping versions available to the general public.
3DMark is a trademark of MadOnion.com

Prior to overclocking, I did let the tests re-loop a few times and there were no issues. 😀

Good catch, how did no one notice this? @DustyShinigami, right now you're overclocking your board. Any stability testing should be done with FSB 100, then work your way up from there. The board supports a maximum FSB of 100, dictated by the chipset specification. Anything above that is overclocking - it's "supported" as in, the board can physically run higher speeds, but it's not guaranteed to be stable.

And the thing is, overclocking the FSB also overclocks the PCI and AGP buses because their clocks are derived from the FSB. AGP cards are often tolerant of this, but PCI devices are much more sensitive, and as you approach 40 MHz from the original 33 MHz, they often stop working.

asdf53 wrote on 2025-10-15, 11:51:

Good catch, how did no one notice this? @DustyShinigami, right now you're overclocking your board. Any stability testing should be done with FSB 100, then work your way up from there. The board supports a maximum FSB of 100, dictated by the chipset specification. Anything above that is overclocking - it's "supported" as in, the board can physically run higher speeds, but it's not guaranteed to be stable.

And the thing is, overclocking the FSB also overclocks the PCI and AGP buses because their clocks are derived from the FSB. AGP cards are often tolerant of this, but PCI devices are much more sensitive, and as you approach 40 MHz from the original 33 MHz, they often stop working.

I see. Okay. I could be wrong, but I'm sure I put all BIOS settings to Factory Defaults at one point, so the CPU was reset back to 466MHz and the FSB to 100. I'm sure I did that and tried the sound card, but I still had the same issue. I'll have a play around and see.

EDIT: Also, I'm able to choose the FSB clock for AGP and PCI, respectively. At default, AGP is 1/1 and PCI is 1/2. Other options are 1/3 and 1/4. And AGP is 1/1 and 2/3.

You guys have hit the nail on the head! 😁 Looks like I was dreaming and hadn't put it to defaults and then tested it. The lowest FSB is 66MHz. The PC booted up correctly - no blank screen - and it's detected the card as a new audio device. 😁 Thank you!

Yes, you need to calculate the AGP and PCI speeds and set the divisor accordingly. The formula is FSB / divisor. For example: FSB 133 and AGP divisor 2/3: (133 * 0.66) = 88 MHz - that's well overclocked, above the spec of 66 MHz. As 2/3 is already the lowest divisor, you cannot run the AGP bus at normal speeds with FSB 133 on this board.

The PCI divisor has more granularity which gives you more flexibility. 1/2, 1/3 and 1/4 should allow pretty much any combination of FSB and PCI speeds with the PCI speed running near 33 MHz.

asdf53 wrote on 2025-10-15, 12:15:

Yes, you need to calculate the AGP and PCI speeds and set the divisor accordingly. The formula is FSB / divisor. For example: FSB 133 and AGP divisor 2/3: (133 * 0.66) = 88 MHz - that's well overclocked, above the spec of 66 MHz. As 2/3 is already the lowest divisor, you cannot run the AGP bus at normal speeds with FSB 133 on this board.

The PCI divisor has more granularity which gives you more flexibility. 1/2, 1/3 and 1/4 should allow pretty much any combination of FSB and PCI speeds with the PCI speed running near 33 MHz.

Uh-oh. Numbers. ^^; I'm trying to process it all, but I'm still a bit confused. I think it's because you say 2/3 is the lowest divisor...? Though I did mention above it has two options - 1/1 and 2/3? Also, how does 2/3 equal 0.66?

The 440BX has serious limitations.
- The RAM frequency is always equal to the system bus (FSB) frequency, i.e. a 1:1 divider.
-- The Northbridge has no dividers for subtiming at 133 MHz, meaning it's 100 MHz memory overclocked to 133 MHz. This requires high-quality memory.
- The AGP frequency can be equal to the FSB frequency (1:1 devider) or differ by a 2:3 divider (the chipset physically doesn't have a 1:2 divider for 133MHz).
- -66/1 = 66 MHz (1:1)
- -100/2/3 = 66 MHz (2:3)
-- 133/2/3 = 89 MHz (2:3)* And that's a very high overclock.
- The PCI bus frequency (the southbridge, which is located below at matherboard) is equal to the FSB frequency divided by :
-- 66/2 = 33 MHz (1:2)
-- 100/3 = 33 MHz (1:3)
-- 133/4 = 33 MHz (1:4) This is the only value of the divider on the motherboard at which everything is fine.

That's all need to know about the 440BX.

If the system is stable at 100 MHz but not at 133 MHz, the problem is either with the RAM or the AGP graphics card.
PCI/ISA devices have nothing to do with it; their clock speeds comply with the standard.
If you select AGP 1:1 and PCI 1:3 with a 133 MHz FSB frequency, you'll get 133 MHz and 44 MHz, respectively.
Everything will magically start glitching, starting with the network card.

DustyShinigami wrote on 2025-10-15, 12:28:

Uh-oh. Numbers. ^^; I'm trying to process it all, but I'm still a bit confused. I think it's because you say 2/3 is the lowest divisor...? Though I did mention above it has two options - 1/1 and 2/3? Also, how does 2/3 equal 0.66?

Forget what I wrote, that was my own math ineptitude coming through there. 1/2, 2/3, 1/3, 1/4 in the settings are multipliers, not divisors, of course.

Here's how it works: You settle on an FSB that you want to use, for example, 100. You want to stay as close as possible to the regular PCI and AGP speeds of 33.3 MHz and 66.6 MHz, so you need to find the correct multiplier setting. You then go through the available multipliers:

For PCI:
FSB 100 * (1/2) = 50 MHz
FSB 100 * (1/3) = 33 MHz
FSB 100 * (1/4) = 25 MHz

For AGP:
FSB 100 * (1/1) = 100 MHz
FSB 100 * (2/3) = 66 MHz

This means, choose 1/3 for PCI and 2/3 for AGP as they result in the correct clock speeds. If none match exactly, choose the one that comes closest. I hope this makes sense.