VOGONS


Turbo XT bus speeds?

Topic actions

First post, by keenmaster486

User metadata
Rank l33t
Rank
l33t

Did Turbo XTs (I understand that this is not a specification, but a colloquialism) generally run the bus at the same speed as the CPU?

If you install one of those "PC Sprint" overclocking boards in your original IBM XT, it runs the CPU at a high clock speed and the rest of the board at 4.77 MHz still... not sure how that is accomplished without additional circuitry to buffer data back and forth, but that's what it does. Is that how "Turbo XT" machines accomplished the "turbo" functionality? Or were they just running the entire system at a faster clock?

World's foremost 486 enjoyer.

Reply 2 of 5, by keenmaster486

User metadata
Rank l33t
Rank
l33t

Did any of them limit the bus to the original 4.77 MHz?

Sounds to me like it's more common to have a Turbo XT that runs the bus and the CPU at either 4.77 MHz (normal) or 8 MHz (turbo) - or possibly a 10 MHz turbo speed if the cards can handle it.

But higher than 8-10 MHz, if you are a manufacturer, that is, would require decoupling from the bus.

World's foremost 486 enjoyer.

Reply 3 of 5, by mkarcher

User metadata
Rank l33t
Rank
l33t

My 10MHz Turbo XT (containing a clone of the DTK PIM-TB10-Z mainboard) runs the ISA bus at 10MHz with an extra wait state (5 clocks/cycle), while the RAM still runs at zero wait states (i.e. 4 clocks/cycle). At 4.77MHz, the ISA bus also uses zero wait states. I obtained these numbers by benching REP STOSW performance on a V20 to either video memory of a fast graphics cards (ISA ATI mach32 and ISA ET4000) or to (conventional) memory. In contrast to the 8088, the V20 is able to saturate the bus on REP STOSW.

Reply 4 of 5, by Jo22

User metadata
Rank l33t++
Rank
l33t++

Hi, I think that a higher bus speed is a good thing. 😁

The 386DX-40 could outperform a 486-25 or 486-33 in terms of throughput thanks to a faster bus speed (25/33 MHz vs 40 MHz) which also was coupled with chipset speed.

For a server that copies data in MByte/s rather than calculating/processing instructions (-where a 486 with cache shines-), this made a difference for the better.

The 8088/8086 Bus Interface Unit (BIU) is very slow,
so perhaps a faster clock speed is almost a requirement to get things moving. 🙂

On the bright side, the older 8-Bit expansion cards do often contain same discrete parts as their 16-Bit versions, so I think they have similar limits.

About 10 MHz should be possible, I suppose. The most critical part would be RAMs, such as the memory on 8088 motherboard.

"Time, it seems, doesn't flow. For some it's fast, for some it's slow.
In what to one race is no time at all, another race can rise and fall..." - The Minstrel

//My video channel//

Reply 5 of 5, by rmay635703

User metadata
Rank Oldbie
Rank
Oldbie
keenmaster486 wrote on 2025-04-14, 17:46:

Did any of them limit the bus to the original 4.77 MHz?

Sounds to me like it's more common to have a Turbo XT that runs the bus and the CPU at either 4.77 MHz (normal) or 8 MHz (turbo) - or possibly a 10 MHz turbo speed if the cards can handle it.

But higher than 8-10 MHz, if you are a manufacturer, that is, would require decoupling from the bus.



“Amstrad clones would drop isa bus to 4.77, It was pretty common for Turbo XT ERSO BIOS to drop the CPU speed to 4.77 MHz when floppy access was being performed, as most used CPU loops for delays. ”

Tandy’s random 1000 series XT class systems ran at speeds up to 9.54 and 10mhz
On the RLX the isa bus was clocked at 8 mhz independently of the cpu using a common crystal , the RL on the other hand had similar behavior to the DTK clone mentioned (from what I remember) inserting a waitstate in certain circumstances.

A few machines would literally change CPU speed to 4.77 on the fly when accessing ISA devices . (Amstrad)

Things were all over the map, someone should test a Juko board and see what they did.