Baoran wrote on 2020-05-27, 09:43:

starcube wrote on 2020-05-27, 01:12:

Baoran wrote on 2020-05-27, 00:12:

I don't know why but I could never manage to match or have faster 486 performance with pci video cards when comparing to my vlb system. If I put my 486 33Mhz cpu to LS-486E motherboard with pci video cards, it always gets worse performance than when I am using the cpu in my normal vlb system. Tried many pci video cards from S3 trio64 and S3 Virge/GX to a pci voodoo 3 card. Many people say that PCI is better but I can't confirm that myself.

I think your 33MHz processor is the real culprit here, bottlenecking the PCI graphics cards. You'd see a difference at 100MHz and above.

Does the cpu affect vidspeed that much? I am getting up to 50% higher numbers in vidspeed with vlb card compared to my pci video cards.
I am getting numbers of over 33k with my vlb card while pci cards are somewhere between 20k and 25k in vidspeed.

Given the processor is actually doing all the drawing, yes. Add the PCI chipset overhead on top and if the CPU is slow, the graphics will lag. VLB makes more sense for a slow CPU like yours as the card is bolted directly to it so there is less bus overhead. PCI really shines with faster processors and leaves VLB in the dust.

Given the above, you should not be testing with vidspeed, because that's just a pure bus bandwidth test, completely ignoring the capabilities of the graphics processor. You should be testing with 3DBench.

hehe always liked those dummy FDD face plates.
But why the 80Mhz on the display?
I'd go with 99 but even that 1Mhz difference annoys me which is why I don't own a case with a display
(it's nothing to do with cost/availability honest)

chinny22 wrote on 2020-05-27, 12:08:

hehe always liked those dummy FDD face plates. But why the 80Mhz on the display? I'd go with 99 but even that 1Mhz differenc […]
Show full quote

hehe always liked those dummy FDD face plates.
But why the 80Mhz on the display?
I'd go with 99 but even that 1Mhz difference annoys me which is why I don't own a case with a display
(it's nothing to do with cost/availability honest)

I know, everyone always asks if it's a real drive 🤣.

I chose 80 because it looks clean. 99 just would not look right no matter what. I had it at 66 before but got tired of it.

I think the display is a quintessential part of the 486 vibe, which is why I love it.

I go with HI and LO when I can't gets me the 'actual' MHz 😜

The MHz display is such a gimmacy load of tosh, I LOVE it!

badmojo wrote on 2020-05-27, 12:15:

I go with HI and LO when I can't gets me the 'actual' MHz 😜

The MHz display is such a gimmacy load of tosh, I LOVE it!

Yup, I don't actually care whether it shows the actual speed, I just like that it's there. 🤣

Intel486dx33 wrote on 2020-05-24, 18:51:

My Pentium 133 build that I am working with right now is a Cadillac compared to the 486. It runs smooth and powerful and cost very little.

But can either of them hold a candle to your.. iPad?

Intel486dx33 wrote on 2020-05-27, 05:53:

It’s like a kit car. Where we both have the same engine. But my kit car is basic model and yours is a sports car model. It really does not make the engine run faster. It just looks after.

That's not how any of this works.

Doornkaat wrote on 2020-05-26, 09:30:

Correct me if I'm wrong but afaik the CPU score is a synthetic score based on CPU calculations alone and not necessarily indicat […]
Show full quote

Correct me if I'm wrong but afaik the CPU score is a synthetic score based on CPU calculations alone and not necessarily indicative of system performance.
Seeing as the overdrive DX4 chips use the same core at the same frequency as regular DX4 CPUs it is to be expected to score the same in this benchmark (within margin of error).
Of course the whole system would perform higher with a 5x86 or similar seeing as the motherboard is capable of running those but I don't think this is the point here, this is not a "fastest Socket 3" deal. Also the same level of performance can be achieved on Socket 5/7 - maybe for a lower price even, but I don't think that's the point either.
I think starcube simply wanted to build a PC that'll get the fastest Intel 486 to perform really well. I think he did a good job at that.👍

You are indeed correct. CPU score does not equal whole-system performance.

System performance is the result of all of its components performing together.

You can have the same CPU score on two systems that actually perform very differently.

starcube wrote on 2020-05-27, 11:26:

I would be very interested in some actual numbers backing this up. Sounds rather implausible given how faster, more capable grap […]
Show full quote

PC-Engineer wrote on 2020-05-27, 05:55:

In 486 world PCI is not in general faster than VLB. Good VLB boards are equal to good PCI boards at same clock. And measuring the Bus-Speed with Speedsys is not the best method to find out the true bus performance.

I would be very interested in some actual numbers backing this up. Sounds rather implausible given how faster, more capable graphics and disk controller chipsets only exist in PCI form. We're not discussing bus speeds here, but whole-system performance.

PC-Engineer wrote on 2020-05-27, 05:55:

In my experiences with comparing L1 cache strategies (using intel/AMD DX2 and DX4 with SIS and UMC chipsets), L1WB makes nearly zero to L1WT in Application-Benchmarks (Doom, Quake, PCP-Bench, Winstone), if you have 256kB or more L2 in WB. Surprisingly, in case of intel DX2 the L1WB/L2WB was a little bit slower than L1WT/L2WB.

I think you shouldn't be mixing Intel and AMD chips in your tests as AMD chips are significantly slower at floating point math than Intel chips at the same clockspeed.

WB L1 on an Intel chip gives you up to a 15% boost in FPU-intensive workloads, so I am actually surprised at your results. Quake especially, since it's so FPU-heavy.

PC-Engineer wrote on 2020-05-27, 05:55:

The 486 WB CPUs from intel where intended for low cost systems without L2 cache.

Where did you read that? That really makes no sense to me. A system with no L2 would be dog-slow regardless of the L1 mode and by the time the DX4 WB appeared (October 1994) cache was cheap.

Having a DX4 WB in a L2-less system would literally make as much sense as having a Ferrari engine in a Fiat 126.

PC-Engineer wrote on 2020-05-27, 05:55:

@starcube
Its a beautiful and well designed 486 system! But why you chose a 486 ALI chipset for a dream system? 😉

Because if you read the datasheets, only the ALi chipset is designed to actually take advantage of EDO DRAM. While late SiS and UMC will boot with EDO DRAM (early revisions will not), they don't actually support EDO data rates and just run it as FPM. 😉

If you like, you can scroll down through the thread within my signature. There i made a comparison with PCI (SIS496 and UMC8881) and VLB (SIS471) boards, a lot of tests with cache behavior of L1/L2 with WB/WT with and without dirty and with different CPUs (P24T, P24D). Additional i made a general comparison with high end 486 CPUs (also AMD SV8B - WT/WB and iDX4 WT/WB) with Quake. L1WB with real 486 CPUs from AMD and intel is overrated.
Maybe you can provide results for Doom, Quake and PCPbench of your system.

That intel intended the 486 WB CPUs to systems without L2 cache i read in a computer magazine (c't). It sounds plausible for me because:
- L1WB gives no benefit together with L2WB (which was very common at the time intel released the WB CPUs)
- Without L2 Cache you gain a benefit of over 15% with L1WB to L1WT (proved in various c't articles)
- the WB CPUs where never delivered to the retail market, only OEM
- intel released the WB CPUs late in 1994 - there the iDX4 was more a like BMW than a Ferrari 😉

And EDO-RAM is cool - but overrated too in systems with L2 Cache.

But the discussion is idle - your intention was to make a clean and cool 486, not the fastest 486.

PC-Engineer wrote on 2020-05-27, 13:11:

There i made a comparison with PCI (SIS496 and UMC8881) and VLB (SIS471) boards

But you have not tested an ALi 1487/1489 board. I recommend this thread: A Tale of Two 486 Motherboards (ALi vs SIS)

PC-Engineer wrote on 2020-05-27, 13:11:

- L1WB gives no benefit together with L2WB (which was very common at the time intel released the WB CPUs)

I remain skeptical of this and similar statements.

PC-Engineer wrote on 2020-05-27, 13:11:

And EDO-RAM is cool - but overrated too in systems with L2 Cache.

What does this mean?

starcube wrote on 2020-05-27, 13:19:

PC-Engineer wrote on 2020-05-27, 13:11:

- L1WB gives no benefit together with L2WB (which was very common at the time intel released the WB CPUs)

I remain skeptical of this and similar statements.

Well, he backed it up with benchmarks, so what do you need to be persuaded?

I will be doing my own testing in the future, when I build my fast 486. I don't have any &EW WB Intel chips but I have these two to compare:

And compare I will.

appiah4 wrote on 2020-05-27, 13:23:

Well, he backed it up with benchmarks, so what do you need to be persuaded?

Because he did not test the chipset I'm using. I recommend this thread: A Tale of Two 486 Motherboards (ALi vs SIS)

starcube wrote on 2020-05-27, 13:19:

PC-Engineer wrote on 2020-05-27, 13:11:

There i made a comparison with PCI (SIS496 and UMC8881) and VLB (SIS471) boards

But you have not tested an ALi 1487/1489 board. I recommend this thread: A Tale of Two 486 Motherboards (ALi vs SIS)

I read it. This article figured out, that the memory and cache performance of ALI is similar to the UMC (feipoa) and EDO RAM performed equal to FPM in his system. If i compare his results with my results i cannot find a proof, that a good configured ALI board performs better than a good SIS or UMC one. And i find no evidence, that PCI is faster than VLB. Sadly i have no ALI for comparison.
Do you have results from your system or comparisons, which can substantiate you claims?

starcube wrote on 2020-05-27, 12:08:

I chose 80 because it looks clean. 99 just would not look right no matter what. I had it at 66 before but got tired of it.
I think the display is a quintessential part of the 486 vibe, which is why I love it.

One option is to find a 2.5 digit display, if your front is exactly the same as the one I have here it should fit out of the box.
The display is a K568:
http://www.minuszerodegrees.net/led_speed_dis … 20-%20photo.jpg

starcube wrote on 2020-05-27, 01:16:

Tetrium wrote on 2020-05-26, 23:49:

The vast majority of Intel DX2 and DX4 chips that I found in the wild (as in dumpsterdiving) were WT. I think I ended up with very few WB Intel DX2/DX4 chips because of that.
There were also several AMD variants of these chips. Some with 8KB cache and some with 16KB and then there were the chips of the other manufacturers. The 486 era was exciting for sure!

I think it's because by the time the WB variants were released in October 1994, Pentium machines were already mainstream.

Iirc Intel never sold the 486DX4 as retail, that's what the OD chips were for. Most people back then were probably not even aware of this difference, they basically only knew stuff like DX2 and DX4.

My guess is that there just aren't that many around, or at least not that ended up inside sold PC systems.

Here in The Netherlands, there didn't appear to be as many s4 Pentium systems either btw, Pentium seemed to start to really take off at around the 430VX or so.

PC-Engineer wrote on 2020-05-27, 05:55:

In 486 world PCI is not in general faster than VLB. Good VLB boards are equal to good PCI boards at same clock. And measuring t […]
Show full quote

In 486 world PCI is not in general faster than VLB. Good VLB boards are equal to good PCI boards at same clock. And measuring the Bus-Speed with Speedsys is not the best method to find out the true bus performance.

In my experiences with comparing L1 cache strategies (using intel/AMD DX2 and DX4 with SIS and UMC chipsets), L1WB makes nearly zero to L1WT in Application-Benchmarks (Doom, Quake, PCP-Bench, Winstone), if you have 256kB or more L2 in WB. Surprisingly, in case of intel DX2 the L1WB/L2WB was a little bit slower than L1WT/L2WB. The 486 WB CPUs from intel where intended for low cost systems without L2 cache.

@starcube
Its a beautiful and well designed 486 system! But why you chose a 486 ALI chipset for a dream system? 😉

These days the main advantage of PCI on a 486 is having access to a much wider array of (usually also much cheaper and easier to find) PCI cards, instead of having to settle for VLB or ISA 😜

Tetrium wrote on 2020-05-27, 14:57:

These days the main advantage of PCI on a 486 is having access to a much wider array of (usually also much cheaper and easier to find) PCI cards, instead of having to settle for VLB or ISA 😜

And I think that's where the real performance advantage of a PCI-based 486 system lies over a VLB one. Especially newer and faster graphics chips are only available as PCI cards.

PARKE wrote on 2020-05-27, 14:00:

One option is to find a 2.5 digit display, if your front is exactly the same as the one I have here it should fit out of the box.
The display is a K568:
http://www.minuszerodegrees.net/led_speed_dis … oto.jpgtest.jpg

This looks wicked cool! Unfortunately it's also pretty much unobtanium. 🙁

starcube wrote on 2020-05-27, 15:36:

This looks wicked cool! Unfortunately it's also pretty much unobtanium. 🙁

I have to admit it took me a year or two to find one. But at least you now know it is an option.

starcube wrote on 2020-05-27, 15:35:

Tetrium wrote on 2020-05-27, 14:57:

These days the main advantage of PCI on a 486 is having access to a much wider array of (usually also much cheaper and easier to find) PCI cards, instead of having to settle for VLB or ISA 😜

And I think that's where the real performance advantage of a PCI-based 486 system lies over a VLB one. Especially newer and faster graphics chips are only available as PCI cards.

My point is that it's easier and cheaper to fit out an existing 486 board if it has PCI slots because the expansion cards will probably be cheaper and you have more choice when it comes to what you want to fit your rig with.
Obviously you can get higher performance stuff like a Voodoo 5 or perhaps a SATA controller or why not a SB Audigy2 with watercooling? The thing is, even if much more modern PCI cards could be fitted, it doesn't guarantee it will increase performance and besides if you really want more performance from a typical 486 machine, you might as well just build a Pentium or something else more recent.
And besides, newer parts don't automatically equal more performance anyway. Not when it comes to 486 machines.

starcube wrote on 2020-05-27, 15:35:

Tetrium wrote on 2020-05-27, 14:57:

These days the main advantage of PCI on a 486 is having access to a much wider array of (usually also much cheaper and easier to find) PCI cards, instead of having to settle for VLB or ISA 😜

And I think that's where the real performance advantage of a PCI-based 486 system lies over a VLB one. Especially newer and faster graphics chips are only available as PCI cards.

So would you say that PCI video cards newer than your Diamond card offer diminishing--if any--gains? I have a Socket 3 board with VLB and PCI and use a Virge 325 2MB just because it is small and has great compatibility. Use DOS 95% of the time.