Intersting...

My SS7 BIOS has a setting called "Cyrix 6x86/MII CPU ID" but this might not be the same thing. It's possible this feature gets auto enabled, the board / BIOS is quite new.

Maybe Cyrix itself had a utility for this? I remember there was a need for a utility to enable internal cache on various Cyrix upgrade chips.

Yeah, the "Cyrix 6x86/MII CPU ID" option I think is to properly detect the CPU, but i'm not sure if it has something to do with the Linear Burst being enabled or not. In nearly all the SS7 boards I came across, as soon as you put a Cyrix MII CPU on them, a new BIOS option would magically appears that let you enable or disable Linear Burst mode. If you don't have a Cyrix installed, you won't be able to see that option. You also won't be seeing it if you're using a chipset that doesn't support Linear Burst mode, like the Intel ones.

I tried all the Cyrix utilities I came across the net, but no one seemed to mess with the Linear Burst setting.
Also, using Sandra, Aida, Astra, Everest, or whatever system information tool comes to mind, no one mentions a bit about Linear Burst being enabled or not.

Using CTCM7, I see a line that says "Cyrix-LRU" on the "Write Strategy L1" line, but i'm not sure what that really means. In my Gigabyte 5SMM board that line switches between "Cyrix-LRU" and "Optimized-LRU" on each run without any apparent reason, it doesn't matter if I enable Linear Burst in the BIOS or not. On the PCChips board, it always says "Cyrix-LRU".

Oh I see!

Hmm I actually recently got a IBM 6x86 PR300 which I haven't tested yet...

Time to give it a spin 🤣

Stay tuned

Ok so I gave it a spin!

Set the voltage, FSB and multi according to what's written on the Heat spread thing (the seller removed it).

Booted up right away. I took pics of the post screen and BIOS options. But no "linear mode" option as I can tell.

I ran my usual DOS benchmarks (Norton sysinfo, speedsys and 3dbench 1.0c). Speedsys locks up but the other tests all completed.

I will leave this chip in my system for a while. So if you want anything tested (only got DOS however) let me know.

Nice IBM cpu you got there, maulwurf!

The 2.9V parts do get quite hot. I got myself a couple 2.2v parts which I intend to build 1 or 2 systems around and overclock a bit.
Btw, nasty crack you have there 😜
I suppose that monitor is just for testing anyway, right?
I use a 14 inch CRT which, strangely, has a build year of 2001. It supposedly came from South Afrika 🤣.

Anyway, I've always wondered in how much the Linear Burst setting had any actual performance benefits.

My LCD took a dive recently 😢

I was a bit clumsy when playing with my retro PC.

Glad it still works though...

I only got this CPU to complete my "Cache slowdown table". Actually I might post it here.

As we can see once we disable L1 cache, it comes down to FSB and what chip you have. The 6x86 design is really efficient (very fast at low clocks) and at 100MHz FSB we get ~ 486DX4-75...

Still missing Pentium MMX results, but in no rush at the moment. Got some other things coming up soon 🤣

Have you gotten any of the more exotic chips to test yet? (Like Winchip and Rise).
The interesting thing about Winchip is, it runs at up to 240Mhz on 3.52v! Meaning it can be used in a Socket 5 board 😀
And they should be jumperable to 4x multi by having their multi jumpers float (= not installing them).
I know hardly anything about Rise though. From a brief glance the Winchip and Rise look to have about similar attributes with the Winchip being the easier one to find.

Edit: Does increasing/decreasing the K6's multi with all caches disabled have any impact on performance btw?

Tetrium wrote:

Have you gotten any of the more exotic chips to test yet? (Like Winchip and Rise)

Nope. Just some UMC 486s but can't test them (no 486 hardware)...

Edit: Does increasing/decreasing the K6's multi with all caches disabled have any impact on performance btw?

Nope does bugger all (no difference in Australian 🤣)

Thanks for taking the time to look at this.
What I forgot to say earlier was that that "magically appearing" BIOS option seems to be something that only appears on certain SiS530 based mobos. Don't know about others, but it seems it doesn't appear in yours.

I also discovered that speedsys hangs when "Cyrix 6x86/MII CPU ID" is enabled in BIOS. If you disable it, speedsys runs fine.

Can you post a screen of CTCM7 running ?

The CPU:

The BIOS option:

Boot screen:

Thanks for that BIOS setting tip!

Speedsys now completed without any issues. Funny there is a marking for a 6x86 233 and it lines up to the pixel 🤣

Great chips I found, as long as the game doesn't use any Intel / MMX / floating point tweaks. The 3dbench score is simply amazing.

Also attached are the images requested!

I'm seeing that in your 2nd pic, on the "Write Strategy L1" results, far on the right, it says "Cy...". Could it be "Cyrix-LRU" ? That is what I suspect has something to do with the Linear mode. Maybe that "Cyrix-LRU" means that Linear mode is being used ? Who knows...

Would it be a bother if you run CTCM7 a couple of more times ? The reason I ask is because in my mobos, each time I run CTCM7, the "Write Strategy L1" results seem to give different results. It switches between "Cyrix-LRU" and "Optimal-LRU" each run. Just run it again as soon as it finishes each run. Maybe you're lucky enough to always get the same consistent result.

Yea no worries, testing this stuf is what we do 🤣

I ran it ~ 10 times and got Cyrix-LRU each time...

A quick google tells me that LRU stands for Least Recently Used.

Found this on google:

Bus Interface Unit
The BIU provides the signals and timing required by external circuitry. The 64-bit data bus supports two different burst cycle address sequence modes. The "one-plus-four" burst mode is compatible with the P54C burst order. Operating the CPU in linear burst mode minimizes bus activity and results in higher performance. Linear burst mode is supported in many existing 64-bit chipsets.

So I guess on the modern boards we use this is just enabled by default especially seeing that it's compatible with Intels burst mode...

I have a Cyrix 5x86 in the cpu junk pile that I've spent some serious time with. Twas a curious oddity of a CPU back in time, with all of its crazy not-enabled-by-default features.

Unfortunately some of the features were broken, and that's why they were off. Like branch prediction. It was broken for all but the final chip revision or some such.

If Cyrix CPUs had an FPU that was less 486-class, they'd be more interesting. 😁 I think their CPUs were almost designed for WinBench and WinStone because high scores in them would make sales in the '90s.

Oh well... then LRU has nothing to do with Linear Burst.
At least I think that if my SiS530 boards has options to enable Linear Burst specifically, then they don't enable it by default. Take the Soyo 5SSM for example. It even comes with a jumper to enable it !
Seems yours is working ok with that Ali chipset. As I said before, CTCM7 gaves me different results each time I run it 😖
Oh well, guess there's no easy way to really determine wether is enabled or not. Thanks for your help so far.

As was determined in Cyrix 5x86 Register Enhancements Revealed, Linear Burst Mode had, at best, a 5.9% performance boost, with the average being more like 0.5%. Linear Burst seemed to assist with FPU more than with ALU.

Mau1wurf, I haven't before seen a 6x86/MII in plain ceramic packaging like that; always with either the early Pentium style gold top or with the black heat spreader. I also haven't before seen a socket 7 chip with round corners like that. I wonder if your chip is a rare version.

Similar to this one I have.

feipoa wrote:

As was determined in Cyrix 5x86 Register Enhancements Revealed, Linear Burst Mode had, at best, a 5.9% performance boost, with the average being more like 0.5%. Linear Burst seemed to assist with FPU more than with ALU.

But when you're talking about Super Socket 7 chips that are weaker than Intel at FPU, every little bit extra that you can get out of it will be big help. Finding a motherboard that supports this feature would still be something that would be desirable.

The FIC PA-2007 (VIA VP2, AT) has a jumper to enable linear burst IIRC. Back in the day it was specifically recommended by Tom's HW & Anandtech for the 6x86.