VOGONS


Asus P5A very slow AGP / Video cards

Topic actions

First post, by Andry

User metadata
Rank Newbie
Rank
Newbie

Well,

I'm going crazy with an Asus P5A Super Socket 7 motherboard. I can't get AGP cards works good.
CPU is AMD K6-2 550 MHz, 512MB SDRAM.

I tried a 3DFX Voodoo3 AGP and Matrox G400 AGP on Windows 98se OS.
Both cards works slow. I score about 660 points on 3D Mark 2000.

Both cards score about 1350 points on a Soyo SY-5EHM (Via MVP3 chipset) + K6-2 450MHz + 256MB SDRAM, with the same OS drivers and settings.

I tried on P5A different BIOS version, tons of different BIOS settings, reduced RAM to 128MB, reinstalled clean OS.

Any ideas?

Thank you,

Andrea

Reply 1 of 38, by F2bnp

User metadata
Rank l33t
Rank
l33t

Hello, you're probably running into what seems to be a classic issue with the Aladdin V chipset. Grab the ALi AGP Utility from here and set Write Combining options to "ALL". That should probably fix the issue, but make sure to test with the Voodoo3 first.

This has been repeated multiple times, but the AGP implementations on most of these early non Intel chipsets were pretty mediocre, so they usually cause all sorts of issues. Your Matrox card might behave in odd ways for example, but 3Dfx cards are free from these issues because they don't make use of any of the AGP bus' features. They just use it as a faster PCI bus, 66MHz vs 33MHz.

Reply 2 of 38, by tpowell.ca

User metadata
Rank Member
Rank
Member

Also be sure that you are not using a '+' CPU as it is known to be incompatible with the later boards.
Any speed of K6-2 or K6-III is fine, just not the K6-2+ and K6-III+ as you will get absolutely terrible performance.

Also, it helps to install the ALi chipset drivers, in case that wasn't done.

  • Merlin: MS-4144, AMD5x86-160 32MB, 16GB CF, ZIP100, Orpheus, GUS, S3 VirgeGX 2MB
    Tesla: GA-6BXC, VIA C3 Ezra-T, 256MB, 120GB SATA, YMF744, GUSpnp, Quadro2
    Newton: K6XV3+/66, AMD K6-III+500, 256MB, 32GB SSD, AWE32, Voodoo3

Reply 3 of 38, by dionb

User metadata
Rank l33t++
Rank
l33t++

Those two are possible, but I think there's a much simpler reason: look at the amount of RAM 😮

The Aladdin V chipset can cache 512MB, but only with the revision G chipset, which is used on the P5A revision 1.06 and later. With earlier revisions of the board and chipset, there's a bug meaning it can only cache 128MB regardless of amount of cache on the board. If you install 512MB on one of these boards, you'll only be caching the first 128MB, not the subsequent 384MB. Moreover, DOS/Win98 start using the RAM from the top down, so all OS processes and probably most applications won't be cached by L2. That gives you a massive performance hit.

So: which revision P5A is this?

Edit: just read that you reduced to 128MB. Then this isn't the issue. In that case, a late revision (>1.06) board with G-revision chipset with K6-2+/3+ could be to blame, or the AGP stuff. Do you have the Aladdin V chipset (AGP) drivers installed?

Reply 4 of 38, by The Serpent Rider

User metadata
Rank l33t++
Rank
l33t++

With earlier revisions of the board and chipset, there's a bug meaning it can only cache 128MB regardless of amount of cache on the board

I think that's not the case here. Early revisions had external TAG RAM and later ones had it integrated. That's more similar to the 430VX/TX limitation.

I must be some kind of standard: the anonymous gangbanger of the 21st century.

Reply 5 of 38, by Andry

User metadata
Rank Newbie
Rank
Newbie

Thanks to everyone for the answers.
I just completed some tests, but the issue is still here.

P5A motherboard version is 1.06. ALI AGP driver V1.54e and ALI IDE drivers V3.4 (the latest from Asus website) are installed.
CPU is a standard AMD K6-2 533 MHz (slightly overclocked to 550 using 100 MHz BUS instead 97 MHz), so not PLUS version.
I tried both BIOS 1007.A and 1011.005.

Installing ALI AGP Utilities as suggested by F2bnp made no differences.
I was not able to find "Write Combining options to "ALL"". I tried to enable "Use UC for WC". I also tried to move GAT mode to 1 or 2, enable AGP 2X, and preset "Turbo".
I performed at least 5 test using different settings, but that produced NO differences at all, and I'm still getting around 660 points istead 1350 (or 1700 using a P3 1000 on another PC).

Thank you so much,

Andrea

Reply 6 of 38, by alvaro84

User metadata
Rank Member
Rank
Member
dionb wrote:

The Aladdin V chipset can cache 512MB, but only with the revision G chipset, which is used on the P5A revision 1.06 and later.

....a late revision (>1.06) board with G-revision chipset with K6-2+/3+ could be to blame, or the AGP stuff.

So it basically comes down to early Aladdin V is for K6-2/3+ (can't cache much but works well with '+' CPUs) and rev. G is for vanilla K6-2 for it can cache 512MB? And both versions are good for plain K6-3...?

Shame on us, doomed from the start
May God have mercy on our dirty little hearts

Reply 7 of 38, by F2bnp

User metadata
Rank l33t
Rank
l33t
Andry wrote:
Thanks to everyone for the answers. I just completed some tests, but the issue is still here. […]
Show full quote

Thanks to everyone for the answers.
I just completed some tests, but the issue is still here.

P5A motherboard version is 1.06. ALI AGP driver V1.54e and ALI IDE drivers V3.4 (the latest from Asus website) are installed.
CPU is a standard AMD K6-2 533 MHz (slightly overclocked to 550 using 100 MHz BUS instead 97 MHz), so not PLUS version.
I tried both BIOS 1007.A and 1011.005.

Installing ALI AGP Utilities as suggested by F2bnp made no differences.
I was not able to find "Write Combining options to "ALL"". I tried to enable "Use UC for WC". I also tried to move GAT mode to 1 or 2, enable AGP 2X, and preset "Turbo".
I performed at least 5 test using different settings, but that produced NO differences at all, and I'm still getting around 660 points istead 1350 (or 1700 using a P3 1000 on another PC).

Thank you so much,

Andrea

I don't have my P5A-B set up at this time, but I managed to find this screenshot online:
0u501cd4d0-1e363bb6-6a8863b0.png

If possible, set GAT to 0 and Frame Buffer Size to "ALL". Also, enable Write Allocation without Hole. Once you've done these changes, try benchmarking once again.

Reply 8 of 38, by dionb

User metadata
Rank l33t++
Rank
l33t++
alvaro84 wrote:

[...]

So it basically comes down to early Aladdin V is for K6-2/3+ (can't cache much but works well with '+' CPUs) and rev. G is for vanilla K6-2 for it can cache 512MB?

Yep. Note that the bug on the late revisions is unique to the P5A(-B), other manufacturers' boards with G-revision Aladdin V have no such problems with the K6plus CPUs. So theoretically a late revision Gigabyte GA-5AX with G-rev Aladdin V would be a better choice for all CPUs, but a) GA-5AX tend to be much rarer than P5A boards and b) if you're using a K6plus, the difference between having L3 cache and not having L3 cache is negligible, so an early revision P5A works fine. Mine is a 1.04 😉

And both versions are good for plain K6-3...?

I think so, although I've never had the chance to try it out myself.

Reply 9 of 38, by meljor

User metadata
Rank Oldbie
Rank
Oldbie
dionb wrote:

Those two are possible, but I think there's a much simpler reason: look at the amount of RAM 😮

In that case, a late revision (>1.06) board with G-revision chipset with K6-2+/3+ could be to blame, or the AGP stuff. Do you have the Aladdin V chipset (AGP) drivers installed?

Small ''correction'': 1.05 and 1.06 P5A and P5A-B both have problems with the mobile (+) cpu's, only 1.03 and 1.04 run fine. I have all of them and can also confirm.

To OP: try using the ali 2.13 agp driver, it works very well.

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 10 of 38, by dionb

User metadata
Rank l33t++
Rank
l33t++
meljor wrote:

Small ''correction'': 1.05 and 1.06 P5A and P5A-B both have problems with the mobile (+) cpu's, only 1.03 and 1.04 run fine. I have all of them and can also confirm.

I understood that the 1.05 only had the G-revision chipset on the P5A-B, not the P5A. You have an ATX P5A 1.05 with G-revision?

Reply 11 of 38, by swaaye

User metadata
Rank l33t++
Rank
l33t++

I came across some old recommendations for those ALI AGP settings when I was trying DVD playback on a P5A. These settings are especially beneficial for video playback efficiency. I don't know if they do anything meaningful for other tasks.

Write allocation enabled without holes
K6 features mask to 3
GAT to 1 (not 2)
AGP2X and sidebanding enabled
Write combining enabled

This isn't stable with all video cards. In fact I think it only worked well with Voodoo3. Other AGP cards tend to be unstable on that chipset in general and tweaking for speed isn't a good idea 🤣

Reply 12 of 38, by F2bnp

User metadata
Rank l33t
Rank
l33t
swaaye wrote:
I came across some old recommendations for those ALI AGP settings when I was trying DVD playback on a P5A. These settings are e […]
Show full quote

I came across some old recommendations for those ALI AGP settings when I was trying DVD playback on a P5A. These settings are especially beneficial for video playback efficiency. I don't know if they do anything meaningful for other tasks.

Write allocation enabled without holes
K6 features mask to 3
GAT to 1 (not 2)
AGP2X and sidebanding enabled
Write combining enabled

This isn't stable with all video cards. In fact I think it only worked well with Voodoo3. Other AGP cards tend to be unstable on that chipset in general and tweaking for speed isn't a good idea 🤣

I was looking for these! I remember following them when you first posted them in that thread, but I found that the option that really mattered for 3D performance was the Frame Buffer Size. Default and Disable kill performance, ALL is a must really, I never bothered to try the other ones.

I suppose if you want the ultimate AGP configuration for 3Dfx you have to follow the options you shared, as well as the Frame Buffer Size one. I really think this will fix the OP's issue, as I remember being utterly bamboozled by a GA-5AX board at some point. It ran fine for a while and I took performed some tests and benchmarks. Then, out of the blue, performance went to shit, certainly higher than 50% decrease.
You can actually see it in thiswonderful thread where no one really knew how to help me, as these things were slightly less understood back then, I even reached a wrong verdict 🤣 .

It's only a few months later than that thread when I found the AGP utility thanks to swaaye and realized this is what was holding my system back.

Reply 13 of 38, by meljor

User metadata
Rank Oldbie
Rank
Oldbie
dionb wrote:
meljor wrote:

Small ''correction'': 1.05 and 1.06 P5A and P5A-B both have problems with the mobile (+) cpu's, only 1.03 and 1.04 run fine. I have all of them and can also confirm.

I understood that the 1.05 only had the G-revision chipset on the P5A-B, not the P5A. You have an ATX P5A 1.05 with G-revision?

Sorry, i do not. I have the p5a-b 1.03, 1.04 and 1.05 and the p5a 1.04 and 1.06 (and maybe a 1.03, not sure, think it is a 1.04 also).

Certainly my p5a-b 1.05 is a ''G'' and also my p5a 1.06 is a ''G'' revision. From multiple sources on the net i understood that 1.05 and 1.06 from both boards gave problems. But i don't know about the p5a 1.05 not using a ''G''.

I also have a p5a 1.06 with an ''E'' chipset strangely enough, unfortunately that one doesn't work and i can't seem to find any damage on it.

btw, did you know about the mod here on vogons? I still have not tested this yet but it seems to fix the k6-+ problems of these boards: K6-2/3+ Success on Asus P5A 1.06 - simple mod

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 15 of 38, by meljor

User metadata
Rank Oldbie
Rank
Oldbie
dionb wrote:

I'd seen it, yes. But I only have a 1.04 with E-rev, so no need. My K6-3+ is happy as it is 😀

Yeah, so is mine so there is no real ''need'' other than the fact that i like these mods 😎

But i only have one working 1.05 p5a-b and one working 1.06 p5a so i think i'll wait until i find another one of these to try the mod.

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 16 of 38, by Andry

User metadata
Rank Newbie
Rank
Newbie

Hello to everyone,

I tried to configure ALI AGP Utility as suggested by swaaye and F2bnp.

Advance configuration section of ALI AGP Utility was grey/disabled, so I tried to upgrade ALI AGP Drivers to V2.13 (as suggested by meljor), and this enabled it.

Unfortunately new settings produced no results for me. 3D Mark 2000 score is ever around 660/680 points.

Today later I'll do a final (senseless) try replacing the CPU.

Thank you,

Andrea

Reply 17 of 38, by meljor

User metadata
Rank Oldbie
Rank
Oldbie

Learn something new everyday, didn't know this myself and the p5a series is my favorite....thought only the ''+'' were affected. Found this on the web somewhere, if your test with a slower/older cpu works please confirm. Perhaps the mod i linked to earlier will also fix this problem?

Found this:

it depends on the chip to some extent the first few batchs of 500, 533 and 550 chips will run fine on the any P5A but the newer batch's sence around the time the + line of chips came out do not. I know some people running these combos and there fine I also now some people who tryed these combo's and they did not work ?

Any way this is what I do Know
AMD made a change to the core instruction set to all newer K6 chips that makes them incompatible with the ICC chip on the revision 1.5 and 1.6 P5A and P5A-B motherboards and the 1011_4 beta will show the correct speed and chip and even identifiy the + chips correctly but the problem is not with the bios it's with the Hardware on the board as in the ICC chip

if your getting really low benchmark results as in your system is benching like a P200 then you have this problem and no bios or tweak program is going to fix it I tryed them all on 3 diffurent Asus P5A's rev. 1.6 motherboards with 2 diffurent K6-2-500's and a K6-3+ 450
pop the same chips in a POS Amptron all in wonder, a FIC 503+, and a Acorp Ali5 and they run fine and bench normal put then in a P5A boots fine acts'like it's running fine just kinda slow bench them and it says there running like P100's ?

you can try the beta bios and all the tweak prog's out there but I'd bet it's not going to help.

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 18 of 38, by Andry

User metadata
Rank Newbie
Rank
Newbie
meljor wrote:
Learn something new everyday, didn't know this myself and the p5a series is my favorite....thought only the ''+'' were affected. […]
Show full quote

Learn something new everyday, didn't know this myself and the p5a series is my favorite....thought only the ''+'' were affected. Found this on the web somewhere, if your test with a slower/older cpu works please confirm. Perhaps the mod i linked to earlier will also fix this problem?

Found this:

it depends on the chip to some extent the first few batchs of 500, 533 and 550 chips will run fine on the any P5A but the newer batch's sence around the time the + line of chips came out do not. I know some people running these combos and there fine I also now some people who tryed these combo's and they did not work ?

Any way this is what I do Know
AMD made a change to the core instruction set to all newer K6 chips that makes them incompatible with the ICC chip on the revision 1.5 and 1.6 P5A and P5A-B motherboards and the 1011_4 beta will show the correct speed and chip and even identifiy the + chips correctly but the problem is not with the bios it's with the Hardware on the board as in the ICC chip

if your getting really low benchmark results as in your system is benching like a P200 then you have this problem and no bios or tweak program is going to fix it I tryed them all on 3 diffurent Asus P5A's rev. 1.6 motherboards with 2 diffurent K6-2-500's and a K6-3+ 450
pop the same chips in a POS Amptron all in wonder, a FIC 503+, and a Acorp Ali5 and they run fine and bench normal put then in a P5A boots fine acts'like it's running fine just kinda slow bench them and it says there running like P100's ?

you can try the beta bios and all the tweak prog's out there but I'd bet it's not going to help.

Immediately after reading your post I remembered that I read somewhere that ALL AMD K6 series CPUs (2, 3, +) produced after 03/2000 are not suitable for Asus P5A 1.06 or later.

However, I think that this info should be revised, because my K6-2 533 MHz is from 5th week of 2000 (0005 datecode). I supposed 03/2000 intend March 2000, so from 9th week. If it means from 3rd week, it seems to be OK (do I own one from first P5A 10.6 incompatible CPUs batch ??? 😵 )

I replaced the CPU by a 400 MHz one (year 1999) and the bench is around 1300, as expected.

I think I'd search a 1999 made K6-2 500 - 533 - 550 CPU.. 😀

Thanks to everyone for your help!

Andrea