VOGONS


Reply 20 of 35, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++
Tetrium wrote:

If someone selling an item I pretty much must have, I'll browse his other items to see if theres any €1 bids I can place on items that are nice but not worth shipping on it's own 😉

Hmm that is a good strategy! I shall adopt it 😁

There was another seller with some Cyrix cpus. Starting bid was low, but $10 for shipping. He told me he wanted another 5 bucks for every extra cpu 😖

That was locally in Australia. So I went with the other auction instead.

Reply 21 of 35, by retro games 100

User metadata
Rank l33t
Rank
l33t
Mau1wurf1977 wrote:
Bit of an update. A fellow member of overclockers australia did some L1 and L2 cache testing on a Pentium 200MMX and things look […]
Show full quote

Bit of an update. A fellow member of overclockers australia did some L1 and L2 cache testing on a Pentium 200MMX and things look really good!

With both caches disabled he gets a system information score of 11.03!

Can't wait to get my Socket 7 stuff and see what can be done. It looks like this is more sutable than P3 systems as a P3 system is either super fast or slow as a dog wheras Socket 7 is more granular and you also have a wide range of cpus and clock speeds and you can change the multi...

Well what a good way to start the day!

Were your disabled cache Pentium 3 tests not successful? Were you not able to reduce its speed, so it behaved like a 386?

Reply 22 of 35, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

In terms of being able to reach 386 speed, yes the P3 solution does work.

However I feel that with the Socket 7 solution there are quite a few benefits.

I wasn't able to reach the speed of a faster 386 or a slower 486. The disabled cache just cripples the performance to much. I also found that most P3s and Celerons don't have an unlocked multi, meaning there is only so much flexibility. When I started this project I never thought that you need a 1 GHz chip to reach 386 speeds. I was quite shocked by the result because I know that a 486DX2 66 with cache disabled gives you a perfect 386. So I guess the architectures are just way to different and it seems the closer you are to 386 hardware in terms of age, the easier it is to use the L1 cache trick 😀

Initial tests on the Super Socket 7 platform has shown me that you can easily reach the speed of a 386DX 40 and go even faster. Changing the multi is very easy on these boards as well and there is a wide range of cpus from a Pentium 133 to K6 550 and a few Cyrix chips.

Most of the benefits remain:

Easy to obtain
Cheap
BIOS, Drivers, Manuals directly on the manufacturer website
ATX compatible
Button type batteries
2-3 ISA slots on pretty much every board
AGP slot

So in a nutshell, yes the P3 project was a success but I believe the Super Socket 7 board will give me better results and greater flexibility, being able to cover a much wider "target speed" range...

On Super Socket 7 boards, tests have also shown that L2 cache does have an impact (L2 cache had zero impact on the L3 platform.

Armed with tweaking options for L1 and L2 cache, unlocked multipliers and a wide range of cpus, this project should become a huge success!

EDIT: I also ordered that ATX to AT adapter with switch. You posted it a while ago with a link on ebay. If that gadget works as advertised I might soon be playing with real 386 / 486 hardware as well 😀

EDIT EDIT: And it's about the journey not the destination 😁

Reply 24 of 35, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++
cskamacska wrote:

Epox MVP3-G2/G5 and MVP4A über alles! 😎 No, really, the best.

Yes they are excellent though I prefer others to outbid themselves over these "sought after items"...

I got my boards for just a few bucks and I am sure they are fine. This quest for the "best board" is often totally overrated...

Reply 25 of 35, by retro games 100

User metadata
Rank l33t
Rank
l33t
Mau1wurf1977 wrote:

Armed with tweaking options for L1 and L2 cache, unlocked multipliers and a wide range of cpus, this project should become a huge success!

Excellent. Please post back with the results. Thanks.
Have you also acquired some 386 and 486 mobos?

Reply 26 of 35, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

Not yet. I don't like doing to many things at once 😀 There is plenty of time...

My website with reviews, demos, drivers, tutorials and more...
My YouTube channel

Reply 27 of 35, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

Ok played around with the boards. This is what I found so far:

Aopen board:

- Odd layout. The power and floppy ports are at odd locations
- (only) 2 ISA slots
- BIOS doesn't have an option to disable L1 Cache
- Great Voltage options
- Great support through Aopen website (Bios, manual, jumper sheet)
- Compatible with K6-2+ CPU
- 1024k L2 cache

Iwill board:

- Great layout
- 3 ISA slots
- No support anymore
- No K6-2+ support (although a newer BIOS might fix this, but I can't find one)
- (only) 512k of L2 cache

I got my K6-2+ 550 running on the Aopen board. Quite a fast machine for a Socket 7 board!

k62plus550.png

Reply 29 of 35, by swaaye

User metadata
Rank l33t++
Rank
l33t++

The mobo L2 barely affects performance with a K6+/K6-III chip. The CPU's main L2 fulfills most of its needs. Some people just disable the mobo L2 in order to gain access to faster FSBs that the mobo L2 can't handle.

Besides, the L2 cache on a Socket 7 mobo is really slow. It shares the FSB with everything else, unlike the L2 cache inside a K6+/K6-III which has its own dedicated data bus (like PPro and later). It runs at a tiny fraction of the CPU clock too.

Check out the result in his bench. The CPU L2 is showing about 7x faster than the mobo L2. 😁 Yikes huh?

Reply 30 of 35, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

Yea the on-chip L2 cache surely is impressive...

I tucked the board + cpu away. Maybe I will find a use for it one day.

My website with reviews, demos, drivers, tutorials and more...
My YouTube channel

Reply 31 of 35, by Amigaz

User metadata
Rank Oldbie
Rank
Oldbie
Mau1wurf1977 wrote:
Ok played around with the boards. This is what I found so far: […]
Show full quote

Ok played around with the boards. This is what I found so far:

Aopen board:

- Odd layout. The power and floppy ports are at odd locations
- (only) 2 ISA slots
- BIOS doesn't have an option to disable L1 Cache
- Great Voltage options
- Great support through Aopen website (Bios, manual, jumper sheet)
- Compatible with K6-2+ CPU
- 1024k L2 cache

Iwill board:

- Great layout
- 3 ISA slots
- No support anymore
- No K6-2+ support (although a newer BIOS might fix this, but I can't find one)
- (only) 512k of L2 cache

I got my K6-2+ 550 running on the Aopen board. Quite a fast machine for a Socket 7 board!

k62plus550.png

Here's what a K6-III+ 500mhz result (from my now retired rig) looked like:

90e97a543b.jpg

Seems like you have a later chipset version with more "boost" since your score is higher even though your cpu only has another 50mhz

My retro computer stuff: https://lychee.jjserver.net/#16136303902327

Reply 32 of 35, by 5u3

User metadata
Rank Oldbie
Rank
Oldbie

No, it rather seems that a few MHz more do noticeably affect the SpeedSys CPU score. Here is my system at 500, 550 and 600 MHz:

SSTK63500.PNG
SSTK63550.PNG
SSTK63600.PNG

In real-world applications, you won't notice much of a difference though,

Reply 33 of 35, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

YES! Success!

I just got the K6-2+ 550 to work on my XA100 PLUS!

I found the BIOS here: http://members.driverguide.com/driver/detail. … driverid=233949

I had to pay a few bucks to get the download, but it all worked out.

Initially I couldn't flash it. Flash type unknown. I used awdflash from the Iwill CD but no go. It would save the old bios just fine however...

So I did a RTFM and look at that this board has a BIOS setting for flash protecting the BIOS.

So I changed that and voila it flashed, popped the K6-2+ in, set the Voltage to 2.1 and it boots just fine!

Running some benchmarks now!

I sent Jan Steunebrink an email with the awardflash and bios file attached, hoping he might be interested in hosting it!

There you go. It pays not to give up right away...

Here the results:

iwillk6.png

EDIT: Here the Aopen board with VIA chipset for comparison (Bugger all difference...):

k62plus550.png

Reply 35 of 35, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

The BIOS?

Of course!

Attachments

  • Filename
    XA100PLUS1027.zip
    File size
    129.54 KiB
    Downloads
    345 downloads
    File comment
    Final BIOS for Iwill XA100 Plus
    File license
    Fair use/fair dealing exception
  • Filename
    AX59PRO236.zip
    File size
    313.35 KiB
    Downloads
    358 downloads
    File comment
    Final BIOS for Aopen AX59 Pro
    File license
    Fair use/fair dealing exception