VOGONS


K6-III+ 450 fastest overclock

Topic actions

Reply 40 of 107, by idspispopd

User metadata
Rank Oldbie
Rank
Oldbie

Somehow that wasn't the post I had in mind, but that's not important.
It seems boards with SiS530 as Baby AT exist (Abit SK5, ASUS P5S-B, ASUS P5-99B), but those have integrated video so no AGP slot. Maybe the chipset was introduced too late (1998) to produce many Baby AT models?

Reply 41 of 107, by meljor

User metadata
Rank Oldbie
Rank
Oldbie
Skyscraper wrote:
Im pretty sure I posted benchmarks with a K6-3+ @ 5*120 MHz a month or two ago ;) […]
Show full quote

Im pretty sure I posted benchmarks with a K6-3+ @ 5*120 MHz a month or two ago 😉

The system was very stable with only 2.0V using my GA5-AX but if I used more than one memory stick the performance took a nose dive what ever BIOS settings I used.
With a single 256MB stick the system was really really fast for a socket 7 system. The OS I used was Windows 98SE.

k63plus600120superpi.jpg

k63plus600120sandra9.jpg

k63plus600120Geforce.jpg

Very nice, it beat my p3-450 on asus p3b-f and geforce 2mx400 at 3dmark (my score was 300 points lower). Too bad games don't work as well as 3dmark99 does: my k6-3+ @ 600 gets some good scores too but in actual games the intels are much faster.

p2-450 is slower in 3dmark (no SSE) but beats a k6 easily in games.

Example: p2-450, p3b-f, voodoo3 3dmark99: 2971 3dmarks.
K6-3+ 6x100, p5a, voodoo3 3dmark99: 3766 3dmarks
p2 in bechmark N.I.C.E. 2 (a.k.a breakneck) 55,5fps
k6-3+ in N.I.C.E. 2 43,7fps

K6-3+ 600mhz scored about the same as a 350mhz pentium2 in this game.

asus tx97-e, 233mmx, voodoo1, s3 virge ,sb16
asus p5a, k6-3+ @ 550mhz, voodoo2 12mb sli, gf2 gts, awe32
asus p3b-f, p3-700, voodoo3 3500TV agp, awe64
asus tusl2-c, p3-S 1,4ghz, voodoo5 5500, live!
asus a7n8x DL, barton cpu, 6800ultra, Voodoo3 pci, audigy1

Reply 42 of 107, by F2bnp

User metadata
Rank l33t
Rank
l33t

If only Skyscraper or lazibayer could give us some results with a Voodoo 3 3000 on an overclocked FSB, we could actually find out just how much of a difference FSB makes on these CPUs.

Reply 43 of 107, by Mamba

User metadata
Rank Oldbie
Rank
Oldbie

No baby AT with AGP slot was made, if for "baby AT" you mean smaller than a P5A-B.

Reply 44 of 107, by F2bnp

User metadata
Rank l33t
Rank
l33t
Mamba wrote:

No baby AT with AGP slot was made, if for "baby AT" you mean smaller than a P5A-B.

P5A-B is baby AT.

Reply 45 of 107, by sliderider

User metadata
Rank l33t++
Rank
l33t++
Mamba wrote:

No baby AT with AGP slot was made, if for "baby AT" you mean smaller than a P5A-B.

P5A is the full sized motherboard. P5A-B is the small one. The "B" means "baby".

P5A

asusp5a.JPG

Oh, and no AGP slot?

P5A-B with AGP slot

p5ab.jpg

Reply 46 of 107, by Mamba

User metadata
Rank Oldbie
Rank
Oldbie

P5A-B is baby AT.

In fact I said, if he means a motherboard smaller than a P5A-B, that is baby AT.
There are plenty of motherboard with AGP slot that is baby AT, I'm not sure what he means.

Reply 47 of 107, by F2bnp

User metadata
Rank l33t
Rank
l33t

A motherboard with a smaller form factor than a P5A-B, is not Baby AT. A motherboard with the same form factor is Baby AT.

Reply 48 of 107, by ODwilly

User metadata
Rank l33t
Rank
l33t

He was talking about SS7 boards with the SIS 530 chipset w/agp and Baby AT

Main pc: Asus ROG 17. R9 5900HX, RTX 3070m, 16gb ddr4 3200, 1tb NVME.
Retro PC: Soyo P4S Dragon, 3gb ddr 266, 120gb Maxtor, Geforce Fx 5950 Ultra, SB Live! 5.1

Reply 49 of 107, by Mamba

User metadata
Rank Oldbie
Rank
Oldbie
F2bnp wrote:

A motherboard with a smaller form factor than a P5A-B, is not Baby AT. A motherboard with the same form factor is Baby AT.

Mine was an ironical nonsense... Ok I give up...
My english should be really bad.

For the request, I am afraid that no one ever released a sis530 with deactivable onboard vga.
This results in poor memory performance (really poor) and a nightmare to configure.

Reply 50 of 107, by F2bnp

User metadata
Rank l33t
Rank
l33t

Lost in translation I guess 🤣 . It's alright, English isn't my native language either 😀.

Reply 51 of 107, by Putas

User metadata
Rank Oldbie
Rank
Oldbie

I tried K6 III+ ATZ on GA-5AA rev 1.1 with F6 bios, it could do only 110 MHz FSB with cache or 115 without. 115x4.5 at 2 Volts and barely aligned cooler seems stable.

Reply 52 of 107, by Skyscraper

User metadata
Rank l33t
Rank
l33t

I will dig out my Gigabyte GA5-AX and some K6-3+ CPUs later. I bought 4 K6-3+ 400ATZ a year ago but I have only tested one. As the first CPU I tried did 600MHz stable with 2.0V I diddnt bother with the other ones. It did 4.5*135 for 607 Mhz so it could run high FSB. The old 2.5GB HDD I used got really hot with 45 MHz PCI frequency so I backed down to 5*120MHz where I did most of the benching and game testing. I did not save all sceenshots and results because of laziness but I shall document my findings better this time.

I will have to build some sort of PII system to get some numbers for comparesion, probably a PII 400.

New PC: i9 12900K @5GHz all cores @1.2v. MSI PRO Z690-A. 32GB DDR4 3600 CL14. 3070Ti.
Old PC: Dual Xeon X5690@4.6GHz, EVGA SR-2, 48GB DDR3R@2000MHz, Intel X25-M. GTX 980ti.
Older PC: K6-3+ 400@600MHz, PC-Chips M577, 256MB SDRAM, AWE64, Voodoo Banshee.

Reply 53 of 107, by kanecvr

User metadata
Rank Oldbie
Rank
Oldbie

I'm running my K6-III+ 400ATZ (1.6v) at 600MHz / 2.1v on one of my Lucky Tech P5-MVP3 boards. The thing gives my PIII-500 equipped Vectra SF a run for it's money. About the same performance in games, faster in cpu synthetics and better overall feel. Both systems run creative Voodoo banshees - AGP for the MVP3 (CT6750) and PCI for the Vectra (CT6760) since it doesn't have an AGP slot.

The K6 has the advantage of being able to go slow as well as fast making it a very versatile gaming rig - it can cover games spanning 1989-1999 by simply down-clocking the CPU. Lowest it will go stable is 2.5 x 66MHz. It will post at 1.5 x 66 but it locks up while loading himem.sys for some reason and 2.0 is interpreted as 6x on this CPU. The banshee is great for dos games as well.

I got the CPU recently off ebay - from china - the auction is still up. Planning to buy another one + a slot 1 PIII 800 from the same seller.

Skyscraper wrote:

I will dig out my Gigabyte GA5-AX and some K6-3+ CPUs later. I bought 4 K6-3+ 400ATZ a year ago but I have only tested one. As the first CPU I tried did 600MHz stable with 2.0V I diddnt bother with the other ones. It did 4.5*135 for 607 Mhz so it could run high FSB. The old 2.5GB HDD I used got really hot with 45 MHz PCI frequency so I backed down to 5*120MHz where I did most of the benching and game testing. I did not save all sceenshots and results because of laziness but I shall document my findings better this time.

I will have to build some sort of PII system to get some numbers for comparesion, probably a PII 400.

At 600MHz the K6-3 almost keeps up with a PIII at 500 in games (tested Unreal gold and Quake 3). I'll run and post some benchmark myself if I find the time.

Reply 54 of 107, by carlostex

User metadata
Rank l33t
Rank
l33t
kanecvr wrote:

The K6 has the advantage of being able to go slow as well as fast making it a very versatile gaming rig - it can cover games spanning 1989-1999 by simply down-clocking the CPU. Lowest it will go stable is 2.5 x 66MHz. It will post at 1.5 x 66 but it locks up while loading himem.sys for some reason and 2.0 is interpreted as 6x on this CPU. The banshee is great for dos games as well.

I run it in DOS 2x50MHz and its stable as a rock. How did you make it POST at 1.5 x 66MHz? I believe that the CPU won't interpret a 1.5MHz setting and instead set it as something else. 2x is indeed interpreted by the chip as 6x and 2.5x as 2x.

Reply 55 of 107, by Skyscraper

User metadata
Rank l33t
Rank
l33t
F2bnp wrote:

If only Skyscraper or lazibayer could give us some results with a Voodoo 3 3000 on an overclocked FSB, we could actually find out just how much of a difference FSB makes on these CPUs.

The Gigabyte GA-5AX rev 5.2 is back on the test bench as promised! This time Im going to bench it with my PCI Voodoo 3 "3500" (Its a V3 2000 with a small fan added 😜) and document the results better.

I just need to install Windows 98 and and run some basic benchmarks to see that everything is performing as it should. I noticed that my GA-5AX has the H revision of the ALI Aladdin V chipset, this must be the last revision they made right? I thought revision G fixed the issue with cacheable range but if I rember right my board can only cache 128MB as the earlier revisions, perhaps it has to to with the TAG RAM?

The cachable range wont matter much as I will mostly run with the onboard cache disabled to be able to run high FSB but Im a bit curious about the H revision as I cant seem to find much information about it. Most Gigabyte GA-5AX rev 5.2 seem to use the E revision of the Aladdin 5 chipset.

New PC: i9 12900K @5GHz all cores @1.2v. MSI PRO Z690-A. 32GB DDR4 3600 CL14. 3070Ti.
Old PC: Dual Xeon X5690@4.6GHz, EVGA SR-2, 48GB DDR3R@2000MHz, Intel X25-M. GTX 980ti.
Older PC: K6-3+ 400@600MHz, PC-Chips M577, 256MB SDRAM, AWE64, Voodoo Banshee.

Reply 56 of 107, by Skyscraper

User metadata
Rank l33t
Rank
l33t

Here is a screenshot of my new best SuperPi 1M score with the Gigabyte Ga-5AX and the K6-3+ in Windows 98 SE.

The onboard cache is activated but it can only cache 128MB. Strangly enough I still get better scores with a 256MB stick than with a 128MB stick, I do not really know why.

K63plus5x120GA5AXSup.jpg

When it comes to game performance with my PCI Voodoo III Im still not getting any good results. So far I am on the level of a stock PII 400, in some cases a little slower (OpenGL), in other cases a little faster (Direct 3D). I know the board and CPU can do better as I tested it with both a GF2 MX400 and a GF3 Ti 200 last year and the results were not bad at all, at least not with the GF3 ti 200 but with the PCI Voodoo III an Intel system seems to be the way to go. A Voodoo Banshee is probably a better match for this system when it comes to 3Dfx cards.

Im tempted to test the K6-3+ and PCI Voodoo III with a PC Chips M577 MVP3 board to see if that will produce better results.

Last edited by Skyscraper on 2015-08-02, 20:09. Edited 1 time in total.

New PC: i9 12900K @5GHz all cores @1.2v. MSI PRO Z690-A. 32GB DDR4 3600 CL14. 3070Ti.
Old PC: Dual Xeon X5690@4.6GHz, EVGA SR-2, 48GB DDR3R@2000MHz, Intel X25-M. GTX 980ti.
Older PC: K6-3+ 400@600MHz, PC-Chips M577, 256MB SDRAM, AWE64, Voodoo Banshee.

Reply 57 of 107, by F2bnp

User metadata
Rank l33t
Rank
l33t

How stable is the system? Do you get BSODs or other random hangs with this FSB?

Reply 58 of 107, by Skyscraper

User metadata
Rank l33t
Rank
l33t
F2bnp wrote:

How stable is the system? Do you get BSODs or other random hangs with this FSB?

100% stable at 5x120 even with the motherboard cache activated and all the BIOS settings set to the fastest settings. Not a single crash during days of testing. 2.0V is probably enough but I used 2.1V.

At 5x125 MHz FSB I need to deactivate the motherboard cache or Windows wont load. 5x125 is stable at 2.2V but isnt faster than 5x120 with the motherboard cache activated.

4.5x130 MHz is stable but also slower than 5x120 with motherboard cache.

4.5x135MHz and 4.5x140 MHz isnt stable and wont load Windows other than with a 256MB Infineon stick. With the memory sticks that posted but diddnt want to finish loading Windows starting Windows just took for ever until I hit ctrl+alt+del or powered off. With the stick that loaded Windows I got "Protection error 43" at both these FSB settings when Power Strip loads. I took this as a sign that the PCI Voodoo III diddnt like running at such a high PCI bus speed. Still no blue screens or freezes but I tried SuperPi at 4.5x140Mhz with 2.3V and I got errors. I will try that setting again when its colder as I think I might get a better score than the one I posted above.

I did not try to change the "memory timing" setting from "fast" or use Cas3 at 135 MHz or 140 MHz FSB as I was hunting for a better SuperPi time. Slower memory settings would probably make high FSB more stable but I think the issue with Power Strip and protection error 43 isnt related to the memory but the PCI bus frequency. I will look in to that later when I am finished with the rest of the testing so it wont matter if the HDD gets corrupted.

Im very impressed with this boards stability 😀. Especially the fact that it never freezes even when its not stable, it just wont load Windows, finish 3dmark or produces errors in SuperPi.

Last edited by Skyscraper on 2015-08-03, 06:35. Edited 2 times in total.

New PC: i9 12900K @5GHz all cores @1.2v. MSI PRO Z690-A. 32GB DDR4 3600 CL14. 3070Ti.
Old PC: Dual Xeon X5690@4.6GHz, EVGA SR-2, 48GB DDR3R@2000MHz, Intel X25-M. GTX 980ti.
Older PC: K6-3+ 400@600MHz, PC-Chips M577, 256MB SDRAM, AWE64, Voodoo Banshee.

Reply 59 of 107, by F2bnp

User metadata
Rank l33t
Rank
l33t

Sounds awesome 😀. You say that 4.5x130Mhz is actually slower than 5x120 with L3 Cache, I suppose that's on SuperPi? How about 3D games?

I should buy another K6-III+ CPU and see if it can hit 600Mhz, mine can do 575, but 600 is always unstable 😵 .