New beta Abit bios to fix via 686b timing bug

[The PIT]

Pity Abit, Jetway and Epox are issuing and devoloping bios fixes for people who don't know what they're doing.
Some of these poeple have sat down and devoloping there patches and fixes and have been talking to motherboard manufactures. Via have supposed to have mentioned theres a problem. So if via mention theres a problem it could be that these poeple aren't clueless.
By the way I work around 940 pc's a year hardware wise so I wouldn't say I'm clueless either. I've even beaten a few tech departments. Anyway I'll get this problem licked one way or the other.
Hows the 815 installations going still struggling???

[cbman]

I would oblige you paul but my little brother still has my GF2... (The little leech)

That is ok... the 815 onboard does well enough for what I am using my good computer for right now...

If I get it back in the near future I will up some scores... but for now... I can't help

Charles

------------------
Canadian... Hell Ya!!!

[Rags]

I will be doing some crude bench's this weekend and later this next week, I will post the results.

Rags

[The PIT]

Sorry Technoid I normally don't get grumpy.
Talking of making a balls up got windoze 98 working without a problem (After discovering my dvd had now ceased to play dvds I won't go into that.) then I decided to tinker forgetting the golden rule if it ain't broke don't fix it. Now windoze 98 is on a temporay holiday I'll bring it back screaming from the beach tomorrow. However the fix for windoze 98 didn't work for 2k so it's back for more playing.

[Nuno]

Paul, here´s what I think of Via AGP:

1 - It sucks. I never had troubles with my G400 either on a BX or a AMD750 (gee, now I realize how long I got the G400 ) It always run at 2x, no problem at all. Now I had trouble running it at 2x with KT133 chipsets, either on a Abit KT7 and on a Asus A7V. Latest Asus bios improved it tremendously, but I still get rare system lockups (even on the desktop) that never happens if I force 1x. They are very random, I mean, one time I can loop 3dmark2001 for like 10 hours, the other it crashes on the desktop. I run AGP1x, I care much more for system stability.

2 - About the diferences about 1x and 2x: Generally speaking, very few. Games I can benchmark (Q2, Q3, UT) generally are scoring consistently less 0.5-2 fps less in AGP 1x, a negligible difference. Also for 3dmark´s less that 10-20 points difference. The only noticeable difference I can see between the 2 AGP modes are below:

3dmark2k, texture rendering speed:

AGP 2X: 32Mb - 177 64Mb - 129
AGP 1X: 32Mb - 171 64Mb - 118

Dagooth Moor benchmark:

AGP 2X: 640x480x16: 55.5 1024x768x32: 27
AGP 1X 640x480x16: 47.8 1024x768x32: 22.2

[Technoid]

Sorry to say, but my G400 has never set it's fot on my VIA motherboard.
But it did have a few isues with AGP4 and 32Bit in W2K on my I815 mainboard.

I can't remeber ever seen any benches that makes agp2 or agp4 look as a "quantum leap" over agp1.

Personaly i still think that uping the real MHz on a bus is better than clock doubling!

The only smart thing Intel introduced in the I8xx series was the hub architechure that relived the PCI bus from being the mainboards "spine".

It's pure craze to have a motherboard that is geared towards 1gh+ cpu's that has a central pathway thats on 33MHz.

The best thing would be a AFGP(Advanced Fast Grafics Port) with a mhz speed above 100Mhz

I think that they are opting for clockdoubling because it seems cheaper but stability wise it's stupid!

[Indiana]

I've done some testing with 3DMark2k on my VIA KT133 with the RadeonDDR 64MB:

I ran 3DMark with the AGP4x i normally use, set AGPto 2x (with ZTAGPTool) and reran 3DMark, set AGP1x & reran 3DMark. Then, to be sure I reset AGP4x & ran the bench and set it to 1x & once again tested with 3DMark.

Here are the results:





As you can see the score is getting constantly lower with the lower AGP settings. When resetting to 4x it's still lower than it was the first time with AGP4x (and even AGP2x), but well, MadOnions benchmarks never gave too consistent results and always had the tendency to get slower when consecutively run without reboot - the result of the second AGP1x test is lower than the first AGP1x benchmark as well.

Anyway there IS a notable but not huge difference of 200-300 3DMarks between AGP1x and 4x on the VIA KT133, this is in the same area what I've seen from Intels chips.

P.S.: In contrary to Nuno I don't have problems with AGP4x here, but I had some AGP trouble with my old AMD750-based Mobo (gave me the "bump-back-to-desktop" problem), so this just shows that AGP stability depends on the individual configuration (RAM, Mobo, CPU, gfx-card) and not solely on the northbridge. There are lots of NVidia users that can only run AGP2x or even 1x stable on Intel chips as well.
In general I'd like to second Technoids statement.

[This message has been edited by Indiana (edited 22 April 2001).]

[ALBPM]

Thanks for the tests guys. This has been a good informative thread.

Let's keep this unbiased and collect some more results. I think Rags was going to post some results with an 815 board.

Paul

[The PIT]

Well bugger it still haven't got the problem licked. Dvd's now work, Data corruption gone but still got a problem with the wintv card. I may have solved the freezing with this one only time will tell. However under win98SE the screen (wintv) will still corrupt when running defrag or antivirus in the back ground. Under win 2k this doesn't happen. I'm now yet another card swap position so I may have licked the freezing but only time will tell.
The only differance barring drivers between win2k and win98 is the acpi. Win2k has acpi but win98 doesn't.
The beta bios from Abit has a setting that doesn't do anything pci master time out. Checking this with wpcredit shows the same value whatever you do. I've emailed this fact to abit hopefully they'll do a bit more work.
Dzeus thank you for the option of removing the audio cable but this didn't do anything. However moving the card to pci slot 5 and disabling the highpoint controller has reduced the noise a lot.