The gain of VLB on a 386 is negligible:

As you see the green dot of the 386DX-40 shows a bit lower performance than the 486DX-25 results. Now if you look at the difference at the 486DX-25 between the blue triangles and black dots Local Bus and ISA basically give the same performance. (Small numbers are bus clock)

If it is not for Doom there might be a few applications that are less CPU and more fillrate demanding where the gap will open up a bit.

I have no clue about what chipset that really is and which is why I didn't buy that board, bought the opti board instead.

elianda wrote:

The gain of VLB on a 386 is negligible:

As you see the green dot of the 386DX-40 shows a bit lower performance than the 486DX-25 results. Now if you look at the difference at the 486DX-25 between the blue triangles and black dots Local Bus and ISA basically give the same performance. (Small numbers are bus clock)

If it is not for Doom there might be a few applications that are less CPU and more fillrate demanding where the gap will open up a bit.

I don't understand what you mean... where are 386 VLB results on your graph!? Is this from one motherboard only? Are you comparing a 386 to 486 or what? To me as it looks, there is only one 386 point and that is not even VLB.

To best way to test if there is a gain of VLB on a 386... would be to test the 386 with VLB and then test it without VLB and note the difference between them to see if there is any gain... would it not? Are you trying to beat a 486 with a VLB 386!!?

I don't think you have a valid argument when saying "The gain of VLB on a 386 is negligible" without doing any extensive testing on a 386 with and without VLB... or instead comparing it against a 486! That's just ridiculous.

I have around 18 motherboards which have a 386 and VLB slots, some are doubles so there are more likely to be around 12 different 386 motherboards with VLB. I can tell you, on some of those... there is definitely a gain having VLB.

nforce4max wrote:

I have no clue about what chipset that really is and which is why I didn't buy that board, bought the opti board instead.

Thanks for not buying it. 😁

I have two of those other OPTi boards, but not the EFAR one. It means I now have another different type 386 VLB board to performance test.

From what I gather the EFAR chipset is higher end, mainly used with socket 3, 486's.

How're you getting on with your board? I do like how you have a Mr Bios on it! 😎

The build went well except for my vlb Trident card decided to kick the bucket while I was about to setup my hard drive settings. Yea my board has the Mr Bios installed but to be honest I don't know anything about this but I get the impression that it is very rare.

Now I am off to buy another vlb card 😒

386_junkie wrote:

I don't understand what you mean... where are 386 VLB results on your graph!? Is this from one motherboard only? Are you comparing a 386 to 486 or what? To me as it looks, there is only one 386 point and that is not even VLB.

To best way to test if there is a gain of VLB on a 386... would be to test the 386 with VLB and then test it without VLB and note the difference between them to see if there is any gain... would it not? Are you trying to beat a 486 with a VLB 386!!?

I don't think you have a valid argument when saying "The gain of VLB on a 386 is negligible" without doing any extensive testing on a 386 with and without VLB... or instead comparing it against a 486! That's just ridiculous.

You are right that there is no point with a VLB 386 on the graph. The original intention was to compare the performance of a local bus card to the older ISA cards depending on the CPU. It was part of an earlier thread here at vogons from 2013.
It is my estimation from these results that the 386 will not perform with VLB than with ISA graphics in the Doom bench. You are of course free to not follow the argumentation and provide own results. The bench is the one from Phils ultimate VGA benchmark: Phil's Ultimate VGA Benchmark Database Project

elianda wrote:

386_junkie wrote:

I don't understand what you mean... where are 386 VLB results on your graph!? Is this from one motherboard only? Are you comparing a 386 to 486 or what? To me as it looks, there is only one 386 point and that is not even VLB.

To best way to test if there is a gain of VLB on a 386... would be to test the 386 with VLB and then test it without VLB and note the difference between them to see if there is any gain... would it not? Are you trying to beat a 486 with a VLB 386!!?

I don't think you have a valid argument when saying "The gain of VLB on a 386 is negligible" without doing any extensive testing on a 386 with and without VLB... or instead comparing it against a 486! That's just ridiculous.

You are right that there is no point with a VLB 386 on the graph. The original intention was to compare the performance of a local bus card to the older ISA cards depending on the CPU. It was part of an earlier thread here at vogons from 2013.
It is my estimation from these results that the 386 will not perform with VLB than with ISA graphics in the Doom bench. You are of course free to not follow the argumentation and provide own results. The bench is the one from Phils ultimate VGA benchmark: Phil's Ultimate VGA Benchmark Database Project

Thanks for the link. I do have many benchmarks from earlier tests I could provide.

My major problem with all these projects though is finding the time to do them as I work, and work a lot.

I do have a long standing and ongoing project in the works benching and comparing all the 386 VLB motherboards I do have (I have many)... but I will not have them all done anytime soon... I barely have my weekends, and of the weekends, includes an 8 hour round trip commute with the company car.

For me, to test the benefit of VLB with 386, is keeping as many factors consistent as possible whilst testing the performance on a 386 based system alone, with and without VLB graphics. Testing it against anything else... like a 486 or Pentium, would be like comparing a bicycle with a train or plane.

386_junkie wrote:

After being very quite of late, I came across another 386 motherboard that has VLB. […]
Show full quote

After being very quite of late, I came across another 386 motherboard that has VLB.

I recognise the chipset from a post on here a while back: -

Vogons post; - Help with IBM Mystery Machine

... a picture of that board: - http://i.imgur.com/Ak4ydlu.jpg

That system happened to have a surface mounted DLC3 Blue Lightning and an empty Socket 3.

Does anyone know of this chipset or had any experience with? I've read it may be a re-branded chip though i'm not too sure. The chipset code is the same as on the board I've just received which gets me wondering about 386 performance if it is intended for Socket 3 or 99MHz DLC. 😒

I've recently bought this board but can't find any info on it - manual/jumper settings... With most 386/486 combo boards you can fit the math copro in the 486 socket but at least there were markings on those boards. On this there is only a marking for 486 cpu and nothing about 387. I'd like to test the RapidCAD.

kixs wrote on 2019-08-08, 12:11:

I've recently bought this board but can't find any info on it - manual/jumper settings... With most 386/486 combo boards you can fit the math copro in the 486 socket but at least there were markings on those boards. On this there is only a marking for 486 cpu and nothing about 387. I'd like to test the RapidCAD.

Did you get anywhere with this? I just got a board based on the same chipset (different layout - https://www.ultimateretro.net/en/motherboards/4394), but it refuses to recognize a 387 despite trying four different ones :(

/Eirik