Anonymous Coward wrote:

I always thought the ECS boards were pretty high quality boards back in the 486 days. The UM8810P-AIO has a pretty good reputation as one of the better PCI 486 boards. It's possible that the fake cache chips on your board were put on by the retailer. PCChips on the other hand was/is a horrible company that invented the fake cache scam. I think several years ago ECS and PCChips merged, but they were not the same company when that board was made.

If I weren't so far away I'd be quite happy to take the board off your hands.

hmm....maybe I should give the board a 2nd chance with cache chips from a dead socket 3 mobo I have...

It's odd though that the BIOS post reports the L2 cache, I've upgraded the BIOS to the latest version also..

have you tried memtest86 ?
that one tells you info about the sizes of the L1 and L2 cache as well

Are you sure it's fake cache? Maybe you just have a bad chip or something.

Anonymous Coward wrote:

Are you sure it's fake cache? Maybe you just have a bad chip or something.

Not sure really, wouldn't I get abny error messages ir a non booting computer if a cache chip were broken?

Amigaz wrote:

Not sure really, wouldn't I get abny error messages ir a non booting computer if a cache chip were broken?

Not necessarily. Normally the BIOS shouldn't report L2 cache in case it is broken or missing, but boards with fake L2 cache often have a manipulated BIOS so that they always display some amount of L2 cache.
What happens if you remove the cache chips?

BTW, a PCPlayer benchmark score of 6.4 is about what I would expect in a cacheless board with an unconfigured Cyrix 486. That doesn't explain the abysmal results with the other cards though.

5u3 wrote:

Not necessarily. Normally the BIOS shouldn't report L2 cache in case it is broken or missing, but boards with fake L2 cache ofte […]
Show full quote

Amigaz wrote:

Not sure really, wouldn't I get abny error messages ir a non booting computer if a cache chip were broken?

Not necessarily. Normally the BIOS shouldn't report L2 cache in case it is broken or missing, but boards with fake L2 cache often have a manipulated BIOS so that they always display some amount of L2 cache.
What happens if you remove the cache chips?

BTW, a PCPlayer benchmark score of 6.4 is about what I would expect in a cacheless board with an unconfigured Cyrix 486. That doesn't explain the abysmal results with the other cards though.

I find it odd that they would have tampered with all BIOS versions...even the latest which I'm using
Gonna remove some cache chips and see what happens...if it's boot's I think I have my answer

btw. the CX5x86 GP100 was configured properly...so was the AMD 5x86, I got 0.1-0.2 better score in pcplayer with the AMD 5x86, hehe

The specs on the ECS board is awesome imho..would be really sad if it has fake cache..it's hard to believe it would be the case 😒

the "cachechk" DOS utility specifically tests the performance of various memory accesses per access size. It tries to determine how many cache levels you have and how fast they are. If you run it and it says you only have L1 cache, well that's just not good and probably means either the cache isn't working or it is fake.

BTW, I really like the AMD 5x86 more than Cyrix 5x86. The Cyrix chip isn't really fast unless you can enable its special features, and they frequently aren't stable. Some older steppings of the CPU have bugs with these features too. And the AMD chip's potential 160 MHz clock speed is usually a performance advantage that the Cyrix can't beat.

However, believe it or not but the Cyrix FPU is actually faster than a 486's. So it usually does outperform the Am5x86 in FPU stuff. Unfortunately, that hardly matters because games from 1995 or so were not about FPU at all generally. A Pentium Overdrive FPU will dust that Cyrix FPU though.

I ran a bunch of tests a few years back and the 3 CPUs basically looked like this.

1           Int  FPU
2AMD 160    ++++ ++
3Cyrix 120  +++  +++
4PODP 100   +++  +++++

Combo of synthetic tests in Everest and Sandra, along with playing a little Jedi Knight. The Pentium OD seemed to not work with the L2 cache on the mobo so it was gimped a bit there and the perf was impacted as a result. Cachechk didn't see the L2 with it installed. PODP has a 32KB L1 (16/16) though.

The Pentium OD was tangibly slower for regular Windows stuff than the Am5x86 @ 160. But in 3D games, combined with a Voodoo3, the PODP was obviously the superior CPU. I suppose that considering its lesser integer performance (blaming that L2 issue) that it could be slower in some non 3D games using less FPU.

swaaye wrote:

the "cachechk" DOS utility specifically tests the performance of various memory accesses per access size. It tries to determine […]
Show full quote

the "cachechk" DOS utility specifically tests the performance of various memory accesses per access size. It tries to determine how many cache levels you have and how fast they are. If you run it and it says you only have L1 cache, well that's just not good and probably means either the cache isn't working or it is fake.

BTW, I really like the AMD 5x86 more than Cyrix 5x86. The Cyrix chip isn't really fast unless you can enable its special features, and they frequently aren't stable. Some older steppings of the CPU have bugs with these features too. And the AMD chip's potential 160 MHz clock speed is usually a performance advantage that the Cyrix can't beat.

However, believe it or not but the Cyrix FPU is actually faster than a 486's. So it usually does outperform the Am5x86 in FPU stuff. Unfortunately, that hardly matters because games from 1995 or so were not about FPU at all generally. A Pentium Overdrive FPU will dust that Cyrix FPU though.

I ran a bunch of tests a few years back and the 3 CPUs basically looked like this.

Copy code to clipboard

1           Int  FPU
2AMD 160    ++++ ++
3Cyrix 120  +++  +++
4PODP 100   +++  +++++

Combo of synthetic tests in Everest and Sandra, along with playing a little Jedi Knight. The Pentium OD seemed to not work with the L2 cache on the mobo so it was gimped a bit there and the perf was impacted as a result. Cachechk didn't see the L2 with it installed. PODP has a 32KB L1 (16/16) though.

The Pentium OD was tangibly slower for regular Windows stuff than the Am5x86 @ 160. But in 3D games, combined with a Voodoo3, the PODP was obviously the superior CPU. I suppose that considering its lesser integer performance (blaming that L2 issue) that it could be slower in some non 3D games using less FPU.

Using PC DOS 7 so I don't have cachechk, ran some other cachechk prog I found on the internet http://www.softlookup.com/display.asp?id=9029

I only reported my L1 cache

Would use my Pentium Overdrive's if the fan's weren't as noisy....both of my P OD's are brand new (NOS)

that is the cachechk program i'm referring to 😉

swaaye wrote:

that is the cachechk program i'm referring to 😉

That means I'm in trouble 😲 ( my cache is kaputtski )

Amigaz wrote:

*I can hear a persons voice ringing in my head (he's a member here): real 486 boards use ISA + VLB slots* :wink:

Yeah baby! I don't think the PCI has anything to do with it here. You should get very decent performance out of the board.

If the test programs can not find any cache you can believe them and accept it as the reason the system runs so slow.

The reason putting in a 5x86 improved the system is because the 5x86 has double the L1 cache of any other 486 cpu and thus performance on a 5x86 is much less dependent on L2 cache.

If you want to use another method to check the cache see if it is soldered on the board or if it is in sockets. If it is in sockets crack one open or x-ray it (hehe). Also check the traces going from the cache to the chipset.

I too find it very odd even your new BIOS is tampered with (possibly). That is one consequent scam.

Sure ECS it merged with PC Chips now, but they were a separate company back then. Then again, if they were a reputable company how did they get the idea to merge with PC Chips?

samudra wrote:

... the 5x86 has double the L1 cache of any other 486 cpu ...

Except for the Intel 486DX4. And the Cyrix 5x86... 😉

samudra wrote:

I too find it very odd even your new BIOS is tampered with (possibly). That is one consequent scam.

Agreed. The old PCChips scam usually included a non-flashable BIOS chip soldered onto the board in order to make the scam harder to detect.
Actually I don't think this is one of the "fake cache" boards (unless the cache chips and the BIOS are soldered).

samudra wrote:

Sure ECS it merged with PC Chips now, but they were a separate company back then. Then again, if they were a reputable company how did they get the idea to merge with PC Chips?

PCChips got the idea to merge with ECS, not the other way around: Link

5u3 wrote:

Except for the Intel 486DX4. And the Cyrix 5x86... :wink: […]
Show full quote

samudra wrote:

... the 5x86 has double the L1 cache of any other 486 cpu ...

Except for the Intel 486DX4. And the Cyrix 5x86... 😉

samudra wrote:

I too find it very odd even your new BIOS is tampered with (possibly). That is one consequent scam.

Agreed. The old PCChips scam usually included a non-flashable BIOS chip soldered onto the board in order to make the scam harder to detect.
Actually I don't think this is one of the "fake cache" boards (unless the cache chips and the BIOS are soldered).

samudra wrote:

Sure ECS it merged with PC Chips now, but they were a separate company back then. Then again, if they were a reputable company how did they get the idea to merge with PC Chips?

PCChips got the idea to merge with ECS, not the other way around: Link

Must be broken chip(s) which doesn't cause any error message? 🤣

I'm at a loss here

The chips look real and are in sockets

Have nicked chips from a dead Soyo socket 3 mobo I have which I'm going to try

Do the cache chip traces actually go to the chipset? Some of the fake cache boards had fake traces too that went in loops back to each other 🤣.

5u3 wrote:

samudra wrote:

... the 5x86 has double the L1 cache of any other 486 cpu ...

Except for the Intel 486DX4. And the Cyrix 5x86... :wink:

I stand corrected Sir. I need to brush up on my retro PC facts.

If the chips are in sockets it could have been some kind of swap trick from a shady retailer back in the day, but if this would have been the case he would have also needed to hack the BIOS and hack future BIOS, so we can discard that.

If the BIOS is genuine (and we have no reason to believe it is not) and if the cache is genuine, then it can only be caused by the motherboard. Either it is broken or the jumpers for the cache are set incorrectly. If the latter the BIOS probably will not show any cache present during the POST screen though. What exactly does the POST screen show?

samudra wrote:

I stand corrected Sir. I need to brush up on my retro PC facts. […]
Show full quote

5u3 wrote:

samudra wrote:

... the 5x86 has double the L1 cache of any other 486 cpu ...

Except for the Intel 486DX4. And the Cyrix 5x86... 😉

I stand corrected Sir. I need to brush up on my retro PC facts.

If the chips are in sockets it could have been some kind of swap trick from a shady retailer back in the day, but if this would have been the case he would have also needed to hack the BIOS and hack future BIOS, so we can discard that.

If the BIOS is genuine (and we have no reason to believe it is not) and if the cache is genuine, then it can only be caused by the motherboard. Either it is broken or the jumpers for the cache are set incorrectly. If the latter the BIOS probably will not show any cache present during the POST screen though. What exactly does the POST screen show?

There's no jumper for enabling/disabling the cache
The BIOS post shows a "256kb cache enabled" message

Well, I came across some 486 and even Pentium mobos that had the cache on a separated module, inserted in a special slot. Sometimes those modules just died, and there were no replacements to be found. I have a 486 mobo with a cache module that burned, and the speed dropped just 3-4-5%. BTW, UMC chipset, ISA+VLB+PCI.

fillosaurus wrote:

Well, I came across some 486 and even Pentium mobos that had the cache on a separated module, inserted in a special slot. Sometimes those modules just died, and there were no replacements to be found. I have a 486 mobo with a cache module that burned, and the speed dropped just 3-4-5%. BTW, UMC chipset, ISA+VLB+PCI.

Did it boot with the burned cache and report no errors?

COAST they were called. Cache On A STick. 😀

Some pic's which thickens the weirdness

Post shows my cache