What happens when you press Turbo depends on the motherboard. Since it's probably a board with a fixed clock generator it's probably locked at 33MHz speed-wise. It's an SX so it has no on-die L1 cache. If the motherboard in question does not have L1 cache on board either, the best that it can do is add some wait states, which won't change things all that much. So dropping down to a 386SX20-something is a fairly likely result.

I would suggest using NSSI to benchmark your system with turbo off, that will give you a good indication of where it will land. I would say you'll end up around a 386SX-20, which is roughly a 286-16 so not a bad Turbo state.

appiah4 wrote on 2021-05-18, 17:35:

What happens when you press Turbo depends on the motherboard. Since it's probably a board with a fixed clock generator it's probably locked at 33MHz speed-wise. It's an SX so it has no on-die L1 cache. If the motherboard in question does not have L1 cache on board either, the best that it can do is add some wait states, which won't change things all that much. So dropping down to a 386SX20-something is a fairly likely result.

I would suggest using NSSI to benchmark your system with turbo off, that will give you a good indication of where it will land. I would say you'll end up around a 386SX-20, which is roughly a 286-16 so not a bad Turbo state.

I installed NSSI 060 and it detects my cpu as 33MHz. However I can only use the general information tab and not the menu in order to view details , tools or do benchmarks. ALT button doesn't do anything. Also when I pressed L in order to view information about my sound card I got the DMA error: "EMM386 DMA buffer is too small. Add D=64 parameter and reboot."

tanasen wrote on 2021-05-18, 18:42:

appiah4 wrote on 2021-05-18, 17:35:

What happens when you press Turbo depends on the motherboard. Since it's probably a board with a fixed clock generator it's probably locked at 33MHz speed-wise. It's an SX so it has no on-die L1 cache. If the motherboard in question does not have L1 cache on board either, the best that it can do is add some wait states, which won't change things all that much. So dropping down to a 386SX20-something is a fairly likely result.

I would suggest using NSSI to benchmark your system with turbo off, that will give you a good indication of where it will land. I would say you'll end up around a 386SX-20, which is roughly a 286-16 so not a bad Turbo state.

I installed NSSI 060 and it detects my cpu as 33MHz. However I can only use the general information tab and not the menu in order to view details , tools or do benchmarks. ALT button doesn't do anything. Also when I pressed L in order to view information about my sound card I got the DMA error: "EMM386 DMA buffer is too small. Add D=64 parameter and reboot."

OK, maybe do that and give NSSI another try?

Don't know how. Do I have to edit the autoexec.bat?
Ok, so I figured it must be edited in the config.sys file after the EMM386.EXE . It now reads the Sound Blaster info. However it still doesn't allow me to use the menu.

So I fired up Norton sysinfo again with turbo off and now it detects the speed at 18.5MHz. I guess editing the config.sys file helped. However with turbo on the speed is still 26.6MHz.

Is the program correctly identifying the CPU as 386SX? Because if it gives you the speed rating vs 386DX then all is good, the narrow SX bus causes about 20% performance drop vs DX.
Rather then rely on the calculated speed, which is usually just a loop of slow instructions being timed, run some actual benchmarks and compare results - normal vs de-turbo. This will give you much better idea of the effect, since usually the CPU clock doesn't change, it's either cache being disabled or more wait-states being introduced by the chipset.

Deunan wrote on 2021-05-18, 21:42:

Is the program correctly identifying the CPU as 386SX? Because if it gives you the speed rating vs 386DX then all is good, the narrow SX bus causes about 20% performance drop vs DX.
Rather then rely on the calculated speed, which is usually just a loop of slow instructions being timed, run some actual benchmarks and compare results - normal vs de-turbo. This will give you much better idea of the effect, since usually the CPU clock doesn't change, it's either cache being disabled or more wait-states being introduced by the chipset.

NSSI identified it as an Am386SX 33MHz. Norton systeminfo identifies it as an 80386, 33 MHz and it compares it to a Compaq 386/33MHz with a score of 34.7. Mine (This computer bar) gives 26.6 with turbo and now 9.5MHz with de-turbo. That seems to fit the 20% performance difference vs a DX .

I managed to use the menu in NSSI using a mouse. The CPU Performance benchmark puts my system between an intel 386DX-25 and an intel 386DX-33! With de-turbo it is now below the intel 386DX-25 and over the intel 386SX-16.

Sounds about right.