VOGONS


First post, by jesolo

User metadata
Rank l33t
Rank
l33t

I've been playing around on old PC hardware for a couple of years now.
One thing that I haven't yet fully tested is the turbo switch functionality on these systems.
Please note, the purpose of this topic is not to discuss other methods of slowing down your system (like disabiling the L1 & L2 cache in the BIOS or using slow down utilities like ICD.EXE & ICE.EXE).

I currently have two 486 motherboards, one being an ASUS VL/I-486SV2GX4 motherboard and the other one an ASUS PVI-486SP3.
Both motherboards have an AMD 486DX4-100 NV8T (write-through L1 cache) CPU on board (I also tested the PVI-486SP3 motherboard with an Intel 486DX-33 CPU).
On both systems, the turbo switch (provided of course that it is connected properly to the motherboard) works perfectly and slows down the AMD 486DX4-100 to the speed of an Intel 486DX-33.

A couple of years ago I also owned a Gigabyte GA-5486AL motherboard with an AMD DX4-120 SV8B (write-back L1 cache) CPU on board.
On this particular system, pusing the turbo button did not have any affect on the speed of the CPU.
I also recall having tested the turbo switch functionality on some Pentium 1 boards (that did have the connectors for it) but, also no change in the CPU speed.

What I would like to establish is whether the turbo switch functionality is disabled on CPU's that has a write-back L1 cache scheme?
Or, is there another reason why this would not work on CPU's with write-back L1 cache schemes (maybe it's motherboard specific)?

Reply 1 of 46, by soviet conscript

User metadata
Rank Oldbie
Rank
Oldbie

I recently put together a socket 4 machine running a pentium 66mhz. The motherboard has a connection for a turbo led but no turbo switch. I don't know how much sense that makes. I hooked it up on some off chance but it does nothing.

Reply 2 of 46, by meljor

User metadata
Rank Oldbie
Rank
Oldbie

It seems that there were some early pentiums with a working turbo function (socket 5). Don't know if there were socket 7 boards.

Don't think it is related to the cache but i don't know. Interesting...

asus tx97-e, 233mmx, voodoo1, s3 virge ,sb16
asus p5a, k6-3+ @ 550mhz, voodoo2 12mb sli, gf2 gts, awe32
asus p3b-f, p3-700, voodoo3 3500TV agp, awe64
asus tusl2-c, p3-S 1,4ghz, voodoo5 5500, live!
asus a7n8x DL, barton cpu, 6800ultra, Voodoo3 pci, audigy1

Reply 3 of 46, by j^aws

User metadata
Rank Oldbie
Rank
Oldbie
jesolo wrote:

...
What I would like to establish is whether the turbo switch functionality is disabled on CPU's that has a write-back L1 cache scheme?
Or, is there another reason why this would not work on CPU's with write-back L1 cache schemes (maybe it's motherboard specific)?

The Turbo header working is not a function of L1 Write-back cache because it works on a Pentium P54C - I've got a couple of Socket 7 boards that do this. Moreover, it's a function of chipset and motherboard implementation.

Reply 4 of 46, by jesolo

User metadata
Rank l33t
Rank
l33t
j^aws wrote:
jesolo wrote:

...
What I would like to establish is whether the turbo switch functionality is disabled on CPU's that has a write-back L1 cache scheme?
Or, is there another reason why this would not work on CPU's with write-back L1 cache schemes (maybe it's motherboard specific)?

The Turbo header working is not a function of L1 Write-back cache because it works on a Pentium P54C - I've got a couple of Socket 7 boards that do this. Moreover, it's a function of chipset and motherboard implementation.

Thank you. That is what my initial thoughts were.
I'm thinking of obtaining an AMD Am5x86 133 CPU to see how that will perform in one (or both) of my motherboards.
I'm hoping that the turbo switch functionality will work if I plug in that CPU.

Reply 5 of 46, by HighTreason

User metadata
Rank Oldbie
Rank
Oldbie

I've had some boards where Ctrl-Alt-Num+ and Ctrl-Alt-Num- work but the actual switch wont.

Also, turbo mode can't be toggled in certain CPU modes - such as if a memory manager is loaded.

Turbo can be somewhat cache related as some boards have the option to cause a constant "cache miss" when the system is in deturbo mode.

My Youtube - My Let's Plays - SoundCloud - My FTP (Drivers and more)

Reply 6 of 46, by badmojo

User metadata
Rank l33t
Rank
l33t
HighTreason wrote:

Also, turbo mode can't be toggled in certain CPU modes - such as if a memory manager is loaded.

This isn't always the case in my experience, it depends on the implementation. Some boards actually underclock the CPU, and others mess with the cache, add wait states etc.

I find the whole turbo thing interesting and I'm always careful to correctly set it up on my machines (including the funky MHz display), I've never found turbo mode (or de-turbo I guess) to be useful, particularly for faster CPUs (> 486DX 66). It's just not slow enough, or it's unstable.

Life? Don't talk to me about life.

Reply 7 of 46, by DonutKing

User metadata
Rank Oldbie
Rank
Oldbie

I've had some boards where Ctrl-Alt-Num+ and Ctrl-Alt-Num- work but the actual switch wont.

I've had a couple of 286 boards with both a turbo header, and the ctrl-alt-+/- feature. The boards were referred to as a 'turbo AT' clone system.

Using Norton Sysinfo as a benchmark, the ctrl-alt-minus function seemed to actually drop the speed of the CPU from 10 or 12mhz down to 8MHz - to match the speed of the original AT.
The turbo button seemed to just add wait states to simulate a slow down. Both modes were independent so you could run at 8MHz with additional wait states for a really slow machine.

I personally have never seen any later machine that underclocks the CPU when turbo is disabled - generally they seem to just add wait states.

If you are squeamish, don't prod the beach rubble.

Reply 8 of 46, by HighTreason

User metadata
Rank Oldbie
Rank
Oldbie

I have a 486 who's deturbo can drag the CPU down to about 1.89MHz... Was tempted to try running XP on it to see if it counted as a new record but couldn't be bothered.

Yeah, some later ones actually did nothing at all too and only had the pins there for "case compatibility" - aesthetic reasons.

@badmojo; Correct, I worded my sentence badly and failed to adequately explain things.

My Youtube - My Let's Plays - SoundCloud - My FTP (Drivers and more)

Reply 10 of 46, by jesolo

User metadata
Rank l33t
Rank
l33t
Jolaes76 wrote:

Jesolo ,
I also have the sv2gx4 with an amd 5x86. Turbo switch working here.

That's good to hear. The conclusion I can therefore draw is that the turbo switch functionality has to do with the motherboard and/or chipset and not the actual CPU.

Just as a matter of interest. What CPU speed are you getting (using Norton Sysinfo 8.0 or NSSI 0.6) on your Am5x86 when you de-turbo your system?

Reply 11 of 46, by jesolo

User metadata
Rank l33t
Rank
l33t
badmojo wrote:

I've never found turbo mode (or de-turbo I guess) to be useful, particularly for faster CPUs (> 486DX 66). It's just not slow enough, or it's unstable.

I must admit, based on benchmark tests with Norton Sysinfo 8.0, Landmark 2.0 and NSSI 0.60 (not real world CPU performance), for those CPU's that are able to "de-turbo", the faster the CPU (> 486DX2-66), the faster the CPU is also going to be in it's "de-turbo" mode.
As noted in my original posting, my AMD DX4-100 runs at more or less the equivalent speed of an Intel 486DX-33 in its "de-turbo" mode. However, for older games this could still be too fast (especially if the game was written in the era when your 386DX-25 was the fastest CPU on the market and the 486 wasn't yet available).
Interestingly, my 486DX-33, in its "de-turbo" mode, runs more or less the equivalent speed of an Intel 386DX-33. This should address many of the speed issues experienced in games from that era that wasn't written to take into account the speed of later generation CPU's.

However, and there are already many posts on this (hence why I didn't want to touch on that subject again), by completely disabling your level 1 and/or level 2 cache (or playing around with your wait states and cache cyles in your BIOS), you can slow down a Pentium based CPU to the equivalent speed of a 386 (or even slower) CPU.

Reply 12 of 46, by Jolaes76

User metadata
Rank Oldbie
Rank
Oldbie

Norton Sysinfo 6.01 for my configuration gives

182.3 with turbo on
57.2 with turbo off

"Ita in vita ut in lusu alae pessima iactura arte corrigenda est."

Reply 13 of 46, by Scali

User metadata
Rank l33t
Rank
l33t
j^aws wrote:

Moreover, it's a function of chipset and motherboard implementation.

Indeed, the classic turbo switch just switches the clockspeed. On early 8088 machines you could generally switch it back to 4.77 MHz, so you'd get more or less the same speed as the original IBM PC/XT.
On 286/386/486, it usually switches back to something like 8 MHz, which should make it more compatible with early software.
At some point in the Pentium age, the turbo switch disappeared, as it was no longer relevant.

http://scalibq.wordpress.com/just-keeping-it- … ro-programming/

Reply 14 of 46, by j^aws

User metadata
Rank Oldbie
Rank
Oldbie
jesolo wrote:

...
However, and there are already many posts on this (hence why I didn't want to touch on that subject again), by completely disabling your level 1 and/or level 2 cache (or playing around with your wait states and cache cyles in your BIOS), you can slow down a Pentium based CPU to the equivalent speed of a 386 (or even slower) CPU.

Yep, the Turbo functionality is enhanced by other methods of slowing down, e.g. disabling caches, waitstates, FSB and Multipliers. And with combinations of these, I can take a 200MHz P54C Pentium down to a fast 286, 386, 486 and Pentium - very flexible.

Scali wrote:

...
At some point in the Pentium age, the turbo switch disappeared, as it was no longer relevant.

I've even seen the Turbo Header labelled on a Slot 1 440BX board, unfortunately the board was dead, so couldn't test anything. I doubt it would've functioned anyway.

Reply 15 of 46, by jesolo

User metadata
Rank l33t
Rank
l33t
j^aws wrote:

Yep, the Turbo functionality is enhanced by other methods of slowing down, e.g. disabling caches, waitstates, FSB and Multipliers. And with combinations of these, I can take a 200MHz P54C Pentium down to a fast 286, 386, 486 and Pentium - very flexible.

That is true but, can sometimes become a cumbersome process since it requires a user to either open the case to change jumper settings and/or access the BIOS.

The turbo switch functionality or a software utility (like ICE & ICD) makes the process less cumbersome and risky (especially for a user who might not be that familiar with jumper or BIOS settings).

Reply 16 of 46, by j^aws

User metadata
Rank Oldbie
Rank
Oldbie
jesolo wrote:

...
That is true but, can sometimes become a cumbersome process since it requires a user to either open the case to change jumper settings and/or access the BIOS.
...

Yes, definitely cumbersome with changing jumpers. However, I have plans to build an external jumper switch-box for all controllable jumpers (audio/ video/ motherboard) when everything is finalised, hopefully. Otherwise, I personally wouldn't bother to open the case to change jumpers.

Reply 17 of 46, by jesolo

User metadata
Rank l33t
Rank
l33t
Scali wrote:

Indeed, the classic turbo switch just switches the clockspeed. On early 8088 machines you could generally switch it back to 4.77 MHz, so you'd get more or less the same speed as the original IBM PC/XT.
On 286/386/486, it usually switches back to something like 8 MHz, which should make it more compatible with early software.
At some point in the Pentium age, the turbo switch disappeared, as it was no longer relevant.

Boy, does that bring back old memories.
I remember my dad's first PC (an Olivetti M19 with a green monitor, two 5.25" 360kb floppy drives and 640kb of ram).
It had a turbo switch at the back where you could choose between the standard 4.77 MHz or 8 MHz.
Although 8 MHz is insignificant today, back then it did make quite a difference.
I recall you actually had to switch the PC off, chose your setting and then switch the PC back on (not that we ever did run the PC at 4.77 MHz).

Last edited by jesolo on 2015-02-15, 19:24. Edited 1 time in total.

Reply 18 of 46, by Joey_sw

User metadata
Rank Oldbie
Rank
Oldbie

before clockspeed can be changed in run-time using software,
i always wonderd backthen why older system can change their MHz when the system was on, but newer system can't.
As I did notice some accompanied manual said that dip-switch that control the clockspeed must not be changed when the system was powered on.

-fffuuu

Reply 19 of 46, by j^aws

User metadata
Rank Oldbie
Rank
Oldbie

BTW, one thing the Turbo does as a slowdown method, compared to the cache disabling method, is that it prevents the DOS version of Ultima 7 from speeding the system up by re-enabling L1 cache...