VOGONS


Reply 20 of 47, by feipoa

User metadata
Rank l33t++
Rank
l33t++
BitWrangler wrote:

I thought 256k 512k or a meg didn't matter it was the tag RAM that defined cacheable area.

With direct-mapped cache, normally, you scale the TAG with the cache, so 1024K of cache would use 64kx8 TAG. I'm not really sure what would happen if you ran 256K of cache with a 64kx8 TAG, but I have a feeling that it isn't going to cache all the 256 MB in WT mode.

Ultimate 486 Benchmark | Ultimate 686 Benchmark | Cyrix 5x86 Enhancements | 486 Overkill Graphics | Worlds Fastest 486

Reply 22 of 47, by feipoa

User metadata
Rank l33t++
Rank
l33t++
luckybob wrote:

I have two "high end" 486 boards. Both VIP and UMC 8433 chipsets (iirc), one is a pc chips with no cache and the other has sockets for up to 1mb of cache, but i cannot for the life of me find jumper settings, and there are like 30 jumpers on that board, so just winging it is a non-starter.

my simms are edo/fpm and I never considered even attempting them in a 486.

Both motherboards are in cases and they are currently buried in a pile at the moment. I'd be happy to get them out for you though.

Any luck?

Ultimate 486 Benchmark | Ultimate 686 Benchmark | Cyrix 5x86 Enhancements | 486 Overkill Graphics | Worlds Fastest 486

Reply 25 of 47, by luckybob

User metadata
Rank l33t
Rank
l33t

In my defense, I've had to box 6 things for ebay today, and deal with a drive in my NAS going down. Not to mention getting all the things I needed for apple pie moonshine. ^.^

Both motherboards (pc chips and mitac) only reported 256mb like so:
neZbwXim.jpg

I forgot to take a picture of it, but when I went to 2x sticks, the ram dropped to 128mb

here are the simms i used.
WD4V0AEm.jpg

Either my boards don't support the 128mb per slot, or just not how it is wired on the simm. I know the simms work, eight of them gave me 1gb in my BITCH'N FAST DUAL PENTIUM PRO (tm) setup.

It is a mistake to think you can solve any major problems just with potatoes.

Reply 26 of 47, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Interesting... and thanks for testing this. Are those one sided, or two sided SIMMs?

Sounds like these boards may not act the same as the Biostar MB-8433UUD. When I use one stick of 128 MB FPM, it reads 128 MB. When you use one stick, does the system report 64 MB or 128 MB? I am using double-sided, 16-chip FPM SIMMs. If you have single-sided, 8-chip 128 MB SIMMs, I doubt this configuration would be supported.

I will test this myself soon, but my max configuration would be 128 + 128 + 64 + 64. I am currently held captive by an annoying 386 motherboard repair job.

Ultimate 486 Benchmark | Ultimate 686 Benchmark | Cyrix 5x86 Enhancements | 486 Overkill Graphics | Worlds Fastest 486

Reply 27 of 47, by luckybob

User metadata
Rank l33t
Rank
l33t

2 sided. 16 total chips.

this is the 2nd motherboard i tried it on: https://i.imgur.com/8MNdvdn.jpg

im offering a bounty for proper jumper settings...

It is a mistake to think you can solve any major problems just with potatoes.

Reply 29 of 47, by feipoa

User metadata
Rank l33t++
Rank
l33t++

I haven't seen this particular board before. Does it have a Phoenix BIOS?

OK, 16-chip. Same as me, except the EDO vs. FPM. Curious about a single stick of 128 MB...

Ultimate 486 Benchmark | Ultimate 686 Benchmark | Cyrix 5x86 Enhancements | 486 Overkill Graphics | Worlds Fastest 486

Reply 30 of 47, by luckybob

User metadata
Rank l33t
Rank
l33t

award. its in the previous post. iirc I remember running the part numbers on the ram. I remember reading they were fpm/edo compatible.

i have slept since then, so take it with a grain of salt.

It is a mistake to think you can solve any major problems just with potatoes.

Reply 31 of 47, by feipoa

User metadata
Rank l33t++
Rank
l33t++

I tested more than 256 MB of RAM on a Biostar MB-8433UUD v3.0 and any amount greater than 256 MB does not get reported. Also, two sticks of 128 MB will show up as 256 MB only if the SIMM modules are not placed adjacent to each other. So, only these combinations with modules inserted in: SIMM1 & SIMM3, SIMM1 & SIMM4, or SIMM2 & SIMM4. So if you want all the 256 MB cached, you need to use 1024KB of cache in write-through mode.

I was actually hoping with this thread that there might be some server type systems which sport 512 MB, perhaps those with dual CPU riser cards, or similar.

Ultimate 486 Benchmark | Ultimate 686 Benchmark | Cyrix 5x86 Enhancements | 486 Overkill Graphics | Worlds Fastest 486

Reply 33 of 47, by FFXIhealer

User metadata
Rank Oldbie
Rank
Oldbie

I'm just happy with my 4 sticks of 16MB each in my Pentium Windows 95 build. 64MB of memory in Windows 95 feels like an insane amount - more than games or the system could ever use. Back in the old DOS days of my father's 80486, we first ran with 8MB, then with 16MB. That upgrade felt huge. LOTS of space to play Star Wars: X-Wing, Doom, and Carmen Sandiego games, none of which ever required more than the original 8MB.

292dps.png
3smzsb.png
0fvil8.png
lhbar1.png

Reply 34 of 47, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t
feipoa wrote:
BitWrangler wrote:

I thought 256k 512k or a meg didn't matter it was the tag RAM that defined cacheable area.

With direct-mapped cache, normally, you scale the TAG with the cache, so 1024K of cache would use 64kx8 TAG. I'm not really sure what would happen if you ran 256K of cache with a 64kx8 TAG, but I have a feeling that it isn't going to cache all the 256 MB in WT mode.

I think using a 64kx8 TAG with 256kb cache is possible, but it has to be explicitly supported by the chipset. I think it should be able to cache the full 256MB of RAM, but with a lower hit rate.

I have a 386SX board with 32KB cache. Most boards need 64kb to cache the full 16MB, but this one has some extra tag pieces, and seems to cache all 16MB too.

"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 35 of 47, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Many of the AWARD 486 BIOSes have a TAG width option of 7+1 dirty, or 8+0. At least one manual I read recommends 7+1 when using L2 write-back, and 8+0 for write-through. You think if I use 8+0 with, say, only 512K I can cache all 256 MB?

Ultimate 486 Benchmark | Ultimate 686 Benchmark | Cyrix 5x86 Enhancements | 486 Overkill Graphics | Worlds Fastest 486

Reply 37 of 47, by BitWrangler

User metadata
Rank l33t
Rank
l33t

Heh, SSDs on PCI are probably as fast as most 486 memory access, get up to 133MB/sec vs what was typical 90MB/sec?

Unicorn herding operations are proceeding, but all the totes of hens teeth and barrels of rocking horse poop give them plenty of hiding spots.

Reply 38 of 47, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

I doubt that any PCI card on a 486 can even get close to 133MB/sec. I would guess more like 20MB/sec at best.

"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 39 of 47, by FFXIhealer

User metadata
Rank Oldbie
Rank
Oldbie

I wouldn't know. I've personally never had to use an IDE or SATA riser card via PCI or ISA. All of my computers always had IDE or SATA controllers built into the motherboard.

But on my Windows 98SE system, I did to a Crystal Disk test and I think the board supports ATA/33, but I'm only pulling around 25MB/s read speeds (19.5MB/s write) on my Western Digital Caviar 40GB hard disk drive... for what it's worth. I guess that means the HDD is being limited by the interface. The 3TB Seagate drive running on my 2016 gaming rig is pushing 170MB/s sequential read and 175MB/s write speeds (SATA III, or 6Gbps). And my 850 EVO is pushing 557MB/s read and 534MB/s write on the same bus.

292dps.png
3smzsb.png
0fvil8.png
lhbar1.png