Reply 20 of 38, by feipoa
- Rank
- l33t++
Will it work with the TX486DLC/E? Have you tried it? What is the name of the program? Do you have a link?
Will it work with the TX486DLC/E? Have you tried it? What is the name of the program? Do you have a link?
Nope no links, just a driver disk from the cyrix upgrade package, the "driver" looks specifically for the 486slc , initializes it and patches it's flags on the BIOS everytime you boot it. Until this is done the cpu works and behaves like a normal 386sx.
Your best bet is to look - if it exists ofcourse - a similar program for your specific cpu
ps. I can upload the files if you like but I doubt they will do you any good.
Found it here,
ftp://ftp.eunet.bg/pub/simtelnet/msdos/sysutl/cyrix100.zip
So I am supposedly able enable the L1 cache with this program by Paul Gortmaker. I use the src_ms version with command cyrix -e -f -i1 -r with a Texas Instruments (Cyrix) TX486DLC/E-40GA with markings: 9429FTB6. I assume this is fab date 29th week, 1994.
I get the following results in:
Landmark Benchmark v2.0
before L1 enabled, CPU = 86 Mhz, FPU = 174 Mhz
after L1 enabled, CPU = 129 Mhz, FPU = 186 Mhz
Norton Sysinfo v8.0
before L1 enabled, score = 47.2
after L1 enabled, score = 64.2
Out of curiosity, my AMD 384DX-40 gets,
Landmark CPU = 61.4 Mhz, FPU = 144 Mhz
Sysinfo score = 42.2
I have two questions,
1) Why do cachechk and memspd not show a faster speed for the L1 vs. L2 cache? It appears as if there is no L1 cache still, however the Landmark and sysinfo benchmark scores improved. Speedsys hangs on load when L1 is enabled.
2) This TX486 chip seems to get a lot hotter than the AMD 386DX-40, at least according to my finger. Does it require a heatsink/fan? With just a heatsink, it still gets finger burning hot (>70 C).
Nice boost in performance!
Apart from benchmarks you can also try some demanding games and see if you notice a difference. E.g. Doom or Strike Commander.
Thought I'd finish off this post with a neat little benchmark chart.
Plan your life wisely, you'll be dead before you know it.
Bringing up this thread to share my personal experience with the 486DLC CPUs...
I tested one Cx486DLC-40GP and two TX486DLC/E-40GAs on an ASUS ISA-386C board (the only one I have with a socketed CPU that allows for 40 MHz), using Speedsys 4.78 and Sysinfo 8.0 on a boot floppy disk created in WinXP (so Speedsys reports MS-DOS version 8.00).
With turbo on and cache off Speedsys scores 9.31 and Sysinfo scores 48.6.
With cache on (by same utility that feipoa found and used) Sysinfo scores 65.5. Speedsys shows a pretty garbled screen - CPU score is unreadable, but memory testing graphs don't show anything happening at the 1KB threshold. Moreover, soft-reset (CTRL-ALT-DEL) sends the system into an endless reset loop when loading the OS from the floppy disk.
CPU identification was also fun to watch: Speedsys found a "Cyrix (M5/2-FEh) 13 MHz", Sysinfo got much closer with "Cyrix 486 40 MHz" and Quarterdeck Manifest saw a "Cx486SLC/DLC-45".
Do you know of any 386 board with a socketed CPU that does 40 MHz and can enable the 486DLC cache in the BIOS?
Here's my SpeedSys results. I have an Abit FU340. To date, I haven't seen a 386 with faster Speedsys or cachechk scores. This FU340 seem to be able to handle particulary fast cache and RAM timings.
I haven't seen a socketed 386 motherboard with built-in L1 cache support in the BIOS. I'd be surprised if they exist. Perhaps if a 386 has a flashable BIOS and the board was reputable, maybe the manufacturer released a BIOS update.
I find the software L1 enabler to works well for me. I don't have your hot reboot issues and noticed a substantial performance boost.
Plan your life wisely, you'll be dead before you know it.
wrote:Thought I'd finish off this post with a neat little benchmark chart.
Why is the FPU in the Cyrix chip slower than Intel? I thought in 486, they all were using Intel design under license.
wrote:Why is the FPU in the Cyrix chip slower than Intel? I thought in 486, they all were using Intel design under license.
If you are referring only to the Cyrix FasMath numeric co-processor, I was not aware of any design borrowing from the i387 architecture(seperature 386 NPU). The Intel DX-33 is probably a totally different beast. I think if you compared an i387 (Intel's NPU for a 386) with the Intel DX-33, you'd also find the DX-33 having a greater FPU performance than Intel's own i387. I'd assume this has to do with an improved NPU re-design. Also, the FPU system performance drops quite a bit for 386's due to the NPU being off-chip.
At the time, AMD was known for licensing existing Intel technology time, I think Cyrix was had their own independent designs.
It probably also doesn't help that I have a Cyrix FasMath 25 MHz NPU in a 386 system running at 40 MHz. I've been running it like this for years and haven't had any games crash on me. I actually just noticed this a few months ago as I rarely take the case off. Next time I open the case, I'll switch it with a -40 piece.
Plan your life wisely, you'll be dead before you know it.
I have had several 386 motherboards with BIOS support for L1 cache. I still have two of them. There isn't really anything particularly special about this option though, as it does the same thing as the software utility.
What's up with the memory graph of that 486DLC? It looks like the L1 is disabled.
"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium
I could never understand why Speedsys doesn't indicate 1 KB of L1 cache, we'd need to look at the source code for Speedsys to see if it measures in the 0-1 KB cache range. The benchmark results clearly show that enabling the L1 cache improve results.
Maybe you can test this out with one of your 386's with BIOS L1 support to see if Speedsys responds in the 0-1 KB range. If it does, you may want to try disabling the L1 in the BIOS and enabling it with software to again see what happens with Speedsys in the 0-1 KB region.
Plan your life wisely, you'll be dead before you know it.
Unfortunately my only Cyrix 486DLC chip is permanently soldered to an upgrade module which is meant to go in a 286 (which I do not have here at the moment).
The only chip I have here that will fit into my DLC aware 386 board is a TI 486SXL-40 with 8kb internal cache.
"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium
Hey Guys,
Just got myself a TX486DLC/E-40GA and was wondering if anyone had the cyrix100 cache enabler file? The link feipoa posted is now dead.
Also does anyone know if the TX486DLC will work with a standard 387 FPU or do you need a newer chip?
Thanks!
Vogons Wiki - http://vogonswiki.com
wrote:Hey Guys,
Just got myself a TX486DLC/E-40GA and was wondering if anyone had the cyrix100 cache enabler file? The link feipoa posted is now dead.
ftp://ftp2.freenet.de/pub/.disk0/ftp.simtel.n … tl/cyrix100.zip
In the event this happens to you in the future:
1. Look at link
2. Decipher filename from link (if possible)
3. Google filename. 😀
"I see a little silhouette-o of a man, Scaramouche, Scaramouche, will you
do the Fandango!" - Queen
Stiletto
The TI chip you bought is exactly the same as the Cyrix part. For the most part any FPU will work with it as long as it is rated for the correct speed. I also seem to recall that early model Cyrix FPUs might cause problems with the DLC chips, so in general it's best to stick with late model FPUs from ULSI, IIT or Cyrix.
"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium
What would actually be faster?
Ti486SXL2-50 or Ti486SXL-40
First one is running on 25MHz bus, while the other at 40MHz. I'd assume when running small code to fit its 8kb cache 50MHz would be faster, but otherwise "should" be slower.
Requests are also possible... /msg kixs
I don't know why exactly, but clock doubling in 386 systems doesn't seem to accomplish much. Based on my experiences, I'd rather have the SXL-40.
"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium
Tested with this board: download/file.php?id=89578&mode=view
BIOS supprts L1 & L2: download/file.php?id=100499&mode=view
TI486DLC-40 & ITT4C87DLC-40:
TI486DLC-40 & CYRIX387DX+-40:
Niemand ist nutzlos, er kann immer noch als schlechtes Beispiel dienen...