VOGONS


Post your 386 Speedsys results here

Topic actions

Reply 240 of 325, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

Doesn't cachechk report the amount of cacheable RAM?

"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

Reply 241 of 325, by keropi

User metadata
Rank l33t++
Rank
l33t++

I get these with everything at 0WS

qZKNAkCl.jpg

l5F4amSl.jpg

I don't see me bothering with this mobo anymore, I can't get UMBs to work. Although I set the free regions as R/W Shadow (or vacant) in MR-BIOS , TLB just hangs on initialization just like they are still write-protected or something...

🎵 🎧 PCMIDI MPU , OrpheusII , Action Rewind , Megacard and 🎶GoldLib soundcard website

Reply 243 of 325, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

I think you're right, I confused them.

"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

Reply 244 of 325, by Deksor

User metadata
Rank l33t
Rank
l33t

I'm late to the party, but here's my 386SX25 with VLSI SCAMPSX chipset (from this motherboard 80386 BIOS image collection)

This one actually has a really cool bios that lets you tweak settings like RAM waitstates, by how much you want to divide the CPU speed when turbo is enabled (1,2,3 or 4), and other things.

With the most optimized settings, here's the score I get :

386SX25.gif
Filename
386SX25.gif
File size
14.97 KiB
Views
1722 views
File license
Fair use/fair dealing exception

It scores at 55 in topbench.

I also fixed this motherboard recently which actually uses the exact same chipset, but comes with a SX33 😁

sx33.jpg
Filename
sx33.jpg
File size
86.2 KiB
Views
1722 views
File license
Fair use/fair dealing exception

The BIOS is way less permissive ... But wait a minute, what if I just put the bios from the other motherboard in this one instead ?
You guessed it, it worked flawlessly 😁 (sorry, no clean screenshot this time, it was just a quick test)

sx33bench.jpg
Filename
sx33bench.jpg
File size
460.51 KiB
Views
1722 views
File license
Fair use/fair dealing exception

It scores at 60 in topbench. The smaller gap in topbench is explained by the integrated OAK video card which kinda sucks.

Also, it scores almost equally to my 386DX 40 🤣 (though probably the FPU is making the score of the DX40 way lower than it should)

386DX40.GIF
Filename
386DX40.GIF
File size
8.47 KiB
Views
1722 views
File license
Fair use/fair dealing exception

So, with a good BIOS, the VLSI scampsx chipset seems very cool 😁

Trying to identify old hardware ? Visit The retro web - Project's thread The Retro Web project - a stason.org/TH99 alternative

Reply 245 of 325, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

Your SX doesn't have the FPU installed. The DX does. It skews the speedsys results.

"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

Reply 246 of 325, by Deksor

User metadata
Rank l33t
Rank
l33t

Yeah I noticed that. With a fpu installed, the sx25 falls back to 3.70 or something like that. Which is still nice I guess

Trying to identify old hardware ? Visit The retro web - Project's thread The Retro Web project - a stason.org/TH99 alternative

Reply 248 of 325, by Am386DX-40

User metadata
Rank Member
Rank
Member

Here's my Am386DX-40 on a Seritech SER-386ADIII with 8mb of 70ns ram and a Genoa 7900 (ET4000AX).

Attachments

  • 386DX-2.png
    Filename
    386DX-2.png
    File size
    5.48 KiB
    Views
    1550 views
    File license
    Fair use/fair dealing exception

Reply 249 of 325, by nzoomed

User metadata
Rank Member
Rank
Member

So I just decided to run speedsys on my newly built 386
Ive read that the numeric processor gives you much lower results at around half of what it should be, and even disabling it in the BIOS doesnt seem to fool speedsys into not detecting the NPU.
I guess these results are fairly typical, but my main concern was the linear read speed that seems rather low compared to what others are getting.

It told me I had to disable smartdrive and 386 memory management which I did, my first test with smartdrive enabled did give be a slightly higher linear read speed from the HDD, but nothing to speak of.

Specs:
AMD 386 DX40
ASUS ISA-386U3
Seagate Medallist 520MB HDD
Cheap trident card while testing
16MB 70ns RAM
64KB of cache (soon to upgrade to 256KB)
IIT 3C87-40 NPU

Attachments

Reply 250 of 325, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

You have it backwards. The CPU score with the FPU is what is should be. The score is inflated when the FPU is absent. It's just shitty programming.

Perhaps the memory score is a tad slow. What are you timings set like in the BIOS?

"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

Reply 251 of 325, by nzoomed

User metadata
Rank Member
Rank
Member
Anonymous Coward wrote on 2020-08-07, 05:15:

You have it backwards. The CPU score with the FPU is what is should be. The score is inflated when the FPU is absent. It's just shitty programming.

Perhaps the memory score is a tad slow. What are you timings set like in the BIOS?

Oh, I see.
I was looking at the likes of memory throughput and bandwidth, other sores posted here (without the FPU) appear to be at least double or more of the score I got, even with a DX-25 CPU.
Its too hard to pull out the chip now that its in the case to do another benchmark with it removed.

As far as memory timings, I have no idea what they are, the BIOS has shitty options, I only can set the cache to "fast" or "slow"
Obviously ive got it set to fast, but its only 20ns memory and im hoping the 15ns memory I drop in will be faster.

I hope to experiment with some newer BIOS images, including a MR BIOS ROM that is supposed to have way more options regarding timings, etc.

Last edited by nzoomed on 2020-08-07, 08:56. Edited 1 time in total.

Reply 252 of 325, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

For an Asus board, it has a strangely limited BIOS. MR-BIOS upgrade is a great idea.

Simply dumping in 15ns SRAMs won't change anything. You have to be able to tighten cache timings in order to see a difference. Even then, I find that with 20ns cache I can usually do 2-1-1-1 memory timings on a 40mhz 386. Upgrading probably isn't worth it unless you want to have a dab at 50MHz bus.

"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

Reply 253 of 325, by nzoomed

User metadata
Rank Member
Rank
Member
Anonymous Coward wrote on 2020-08-07, 08:54:

For an Asus board, it has a strangely limited BIOS. MR-BIOS upgrade is a great idea.

Simply dumping in 15ns SRAMs won't change anything. You have to be able to tighten cache timings in order to see a difference. Even then, I find that with 20ns cache I can usually do 2-1-1-1 memory timings on a 40mhz 386. Upgrading probably isn't worth it unless you want to have a dab at 50MHz bus.

OK, well I want to upgrade it to 256KB, and ive got some 15ns chips hanging around, but have plenty of 20ns, if im not going to notice any difference.

Would this affect my memory bandwith results if i can get the timings down?

Reply 254 of 325, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

There are separate adjustments for cache and RAM timings. In theory you should be able to increase bandwidth. On a 40MHz 386 board, about 30MB/sec is normal for cache, and 20MB/sec for DRAM, so you certainly have a little room for improvement.

"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

Reply 255 of 325, by nzoomed

User metadata
Rank Member
Rank
Member

Ok, it will be interesting to see what improvements the other BIOS can provide.

Whats the difference between bandwidth and throughput?
The last results shared before mine say about 60MB/s bandwidth.
Mine is much lower.

Reply 256 of 325, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

I would think bandwidth would be the maximum transfer rate of the bus, and throughput is the actual rate the data moves over the bus. With speedsys, who really knows though.

"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

Reply 257 of 325, by SilverHawk

User metadata
Rank Newbie
Rank
Newbie

PCBell model LA333CSB, NOT PackardBell
TMC Research Corporation PAT38PX v3.10
386DX 33
64KB cache
OPTi 82C392
8MB
Trident TVGA 9000A

Attachments

  • Opti386DX.png
    Filename
    Opti386DX.png
    File size
    20.07 KiB
    Views
    1265 views
    File license
    Fair use/fair dealing exception

Reply 259 of 325, by stalk3r

User metadata
Rank Member
Rank
Member

Ti 486 SXL-40
Ser 386-AD3 (ALi M1429) motherboard
8 MB RAM

Attachments

  • sp.jpg
    Filename
    sp.jpg
    File size
    937.47 KiB
    Views
    1234 views
    File license
    Fair use/fair dealing exception