• Please review our updated Terms and Rules here

Uncached RAM seems very slow (speedsys)

S

smeezekitty

Experienced Member
Joined
Jan 15, 2010
Messages
460
Location
Washington U.S.
I know that the cache has a dramatic effect but these numbers seem dismally low too me.

slow.jpg

Code: 
³ Cache Level 1 ³    31.67 MB/s³    20.57 MB/s³    40.44 MB/s³    30.90 MB/s³ 
³ Cache Level 2 ³    26.35 MB/s³    20.52 MB/s³    22.72 MB/s³    23.20 MB/s³ 
³ Memory        ³    11.32 MB/s³    20.63 MB/s³     6.92 MB/s³    12.96 MB/s³
11.32 and 6.92 seems pretty low. I don't know what is normal though

Also it claims a Y2K bug. Is the detection for it screwed because the BIOS shows 2013 just fine
 
The numbers are not far from what I would expect, but I don't know the exact value you should get either. The values are probably normal, because your CPU speed is on par with the expectations for a 486DX-33 in the picture you posted (between the 386DX-40 and the 486DX2-50). As for the Y2K bug, all my 386-486 boards accept dates above 2000 just fine, but the bug is there. I'm not sure what problems you can have with this bug right now though.
 
Those numbers are about 2/3 of what 486-66 with 256kb of RAM show (only speedsys comparison I could quickly find). Which motherboard and what is the RAM type? Those numbers could be correct if you have a budget board.
 
The numbers are not far from what I would expect, but I don't know the exact value you should get either. The values are probably normal, because your CPU speed is on par with the expectations for a 486DX-33 in the picture you posted (between the 386DX-40 and the 486DX2-50). As for the Y2K bug, all my 386-486 boards accept dates above 2000 just fine, but the bug is there. I'm not sure what problems you can have with this bug right now though.
Click to expand...
The CPU speed seems correct. I just thought the memory might be faster but I guess thats normal.
krebizfan said:
Those numbers are about 2/3 of what 486-66 with 256kb of RAM show (only speedsys comparison I could quickly find). Which motherboard and what is the RAM type? Those numbers could be correct if you have a budget board.
Click to expand...
This is the closest to the board I can find: http://stason.org/TULARC/pc/motherboards/C/COMPUADD-INC-486-486-ISA-HP-SX-DX.html#.UpRiVVNSlhL
It is 80ns parity 30 pin simms.
 
smeezekitty said:
The CPU speed seems correct. I just thought the memory might be faster but I guess thats normal.

This is the closest to the board I can find: http://stason.org/TULARC/pc/motherboards/C/COMPUADD-INC-486-486-ISA-HP-SX-DX.html#.UpRiVVNSlhL
It is 80ns parity 30 pin simms.
Click to expand...

There are surely faster memory chips than 80ns parity. In 1993 my 386DX-40 had 60ns parity. Now I run it with 60ns non-parity. The cache memory also comes in different speeds and can be 15-25ns. My 386 in 1993 had 128K 20ns by default and I've now upgraded it to 256K 15ns with the lowest available timing in BIOS. Playing with cache timing under BIOS I noticed a small, but measurable performance increase.
 
Last edited:
Intel DX4 with 16MB EDO 60ns and 256KB 15ns Cache:

oNeeERsh.png


Same system with AMD 5x86:

UBnL3dnh.png
 
So it seems that a huge plummet is normal.
It does indeed have a problem with Y2k also. I set the date to Dec 31 1999 and it went to 1991 instead of 2000. But once reset it works fine.
 
Ran speedsys on my 386DX40 system.

It has 128KB Cache and 8MB 70ns memory.

Quite a bit faster than your 486...

pMBF3Iq.png
 
Maybe someone can answer this.

I have two 386DX-40 boards with motherboard cache. When disabled they are very very slow.

However I have a 386SX-33 board that has no motherboard cache, yet it is much faster. How is this possible?
 
Mau1wurf1977 said:
Maybe someone can answer this.

I have two 386DX-40 boards with motherboard cache. When disabled they are very very slow.

However I have a 386SX-33 board that has no motherboard cache, yet it is much faster. How is this possible?
Click to expand...

What benchmark numbers do you get?
 
I tried Wing Commander and 3dbench. The SX-33 gets 10 and the cache-less 386DX-40's less than that. And Wing Commander is quite a bit slower also.
 
Some motherboards are just crap. It would help to know which chipsets your board are using. Then you can compare the memory scores to others that use the same board.
 
One has a FOREX and the other a UMC chipset. Both reach around 16 3dbench with cache enabled. But half that score with cache disabled. The 386SX-33 reaches a score of 10 and has no cache on the motherboard.

Unless there is some cache in the chipset or something like that?

Does motherboard cache really double the performance on a 386DX system? Seems a bit too much to me.
 
Last edited:
Cache makes a very large difference. I find that running Linux with mem=16MB to disable use of the uncached RAM, it triples the speed of the system (Linux allocates from top down)
 
Interesting.

So maybe the chipset on the SX-33 board is very good or optimized for good memory performance?

This is the board:

jATh0g4h.jpg
 
Mau1wurf1977 said:
Interesting.

So maybe the chipset on the SX-33 board is very good or optimized for good memory performance?

This is the board:
Click to expand...
I've got this board as well, with an SX-40 CPU but I haven't benchmarked it yet. The truth is that I expected less than 10 3dbench score from a non-cached SX-33. I know that my 386DX-40 without tweaks and with the 20ns 128K cache it had from 1993, was scoring 12.8 in 3dbench. Maybe the DX-line (both chipset & CPU) is designed to work with cache, while the SX without it.
 
Yea I was also surprised. I do use a Tseng ET4000 / Diamond Speedstar 24X cards however. They are pretty fast ISA cards.

For Wing Commander the SX-33 feels pretty much perfect. No scenes that are too fast like chasing that last remaining enemy :) It's a nice little board. If I can get the SX-40 version, or another one of these, I will grab one as these boards seem very reliable and compact. Not much that can fail.

Here some screenshots:

edz94gyh.png


tQrGOMsh.png


PgWeDrmh.png


vPA6UAyh.png
 
Its kind of amazing that a 486 is 2-3 times faster per clock compared to a 386.
That's a vast improvement in architecture.

-edit-
what did you use to make those screenshots?
 
Last edited:
All my gear goes through a KVM and then through a VGA to S-Video capture device. Makes the whole thing very convenient.
 
Made a little video:

http://www.youtube.com/watch?v=lJuFEZ-q_F4&feature=youtu.be

Wing Commander is a game that is very speed sensitive. Is your PC too slow and it will be a slide show, is your PC too fast and you will find it difficult hunting down that last remaining enemy :)

It's a good game to test old retro parts and I like to document the performance so others can use it as a reference. Be it for tuning their emulators or for comparing their retro computer and checking if everything is ok.

The hardware components are shown in the video, so are some common DOS benchmarks.

CPU: AMD 386SX-33
Motherboard: M396F VER 2.2
RAM: 4x1 MB 70ns SIMM
Controller: GoldStar
Storage: 2GB CF card formatted with a single 512MB FAT16 partition
Floppy: USB Floppy simulator
Video: TSENG ET4000
Sound: Sound Blaster Pro 2
OS: MS-DOS 6.22
Mouse: Logitech 7.3 Mouse driver with serial mouse
 
You must log in or register to reply here.
Back
Top