VOGONS


First post, by feipoa

User metadata
Rank l33t++
Rank
l33t++

I put a 50 MHz crystal oscillator in my VLSI-based 386 motherboard and a Texas Instruments 486SXL2-50. These CPUs require software to enable the clock-doubling feature. I run cyrix.exe -cd, which sets bit 6 of CCR0 to enabled (1). You can confirm this by running cyrix.exe -d. You will see that the raw dump shows a value of 40 hex for CCR0, which translates to 1000000 in binary, so bit 6 is indeed enabled. However, chkcpu shows approx 25 Mhz as the CPU speed. The 3dbench scores before and after enabling clock-doubling are the same. Clearly the CPU isn't doubling its clock. I tried running cyrix -cd before and after enabling the L1 cache, but the 3dbench results are the same.

Has anyone had success in enabling clock-doubling on these SXL CPUs and if so how did you do it?

Plan your life wisely, you'll be dead before you know it.

Reply 1 of 5, by feipoa

User metadata
Rank l33t++
Rank
l33t++

I have determined that the AMI Mark V Baby Screamer (VLSI-based) I was using doesn't want to enable clock doubling for whatever reason. A VIA 495-based board I am working with now was able to clock-double the SXL2 just fine using cyrix.exe -cd

Although the bus is set at only 25 MHz to achieve a 50 MHz CPU speed, the results are far more impressive than I first thought. I will report on this later, but the gist of it is that main memory read speed is in consequential to that of raw CPU speed when it comes to these PGA-132 CPUs.

Plan your life wisely, you'll be dead before you know it.

Reply 2 of 5, by Marco

User metadata
Rank Oldbie
Rank
Oldbie

Hello feipoa, all,

Possibly a double post:

In my Morse 386sx / 25 board the systems crashes when -cd‘ing. I tried all possible programs and all your parameters. They all work fine but when double clocking system crashes.

Any idea what might be the issue? 50mhz cpu osci, 32mhz isa osci which runs /4 hard coded giving compliant 8mhz bus.

I am really out of any idea. Thanks for your help.

1) VLSI SCAMP 311 | 386SX25@TI486SXLC2-50@60 | 16MB | CL-GD5428 | CT2830| SCC-1 | MT32 | Fast-SCSI AHA 1542CF + BlueSCSI v2/15k U320
2) SIS486 | 486DX/2 66(@80) | 32MB | TGUI9440 | LAPC-I

Reply 3 of 5, by MikeSG

User metadata
Rank Member
Rank
Member

Did you solder it using a hot air gun or soldering iron?

I believe CD'd doesn't work in the 386sx (TI486SXLC2) when you use a hot air gun to solder the CPU.

The max temp in the manual for the TX486SXLC2-G50-WN (most common version available) is 85C. It's unclear if this includes non-operating conditions, but I've seen multiple people say they have CD issues and they used a hot air gun.

What I do is use a flat soldering tip, lots of flux and basically brush along the pins, never spending more than one second on a single pin.

~

That said, I don't Clock Double my 386sx/TI486SXL2C board because it doesn't add to performance in anything outside of 3DBench. It would add a lot to performance in cache based systems though.

Reply 4 of 5, by Marco

User metadata
Rank Oldbie
Rank
Oldbie

Thanks for that and yes I assume this could be a possible root cause. Although I also used your way of soldering it could be that there is simply a bad connection issue which only while cd‘ing appears.

I will check in the second board as well and report. Thanks again

1) VLSI SCAMP 311 | 386SX25@TI486SXLC2-50@60 | 16MB | CL-GD5428 | CT2830| SCC-1 | MT32 | Fast-SCSI AHA 1542CF + BlueSCSI v2/15k U320
2) SIS486 | 486DX/2 66(@80) | 32MB | TGUI9440 | LAPC-I

Reply 5 of 5, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Do a gentle poke test on the pins with a metal poker, e.g. a dental pick or toothpick. Also check for conductivity of each pin using the multimeter.

Plan your life wisely, you'll be dead before you know it.