WTF ! PCI/AGP 42/84 !

[SidVicious]

I'm using a Duron Morgan AHHAA 1GHz @ 166 x 6 on a A7V333-R BIOS 1014. When I ran Sandra, I noticed that my PCI and AGP busses were WAY out of spec ( 42/84 ). The A7V333 is supposed to support a 166MHz FSB with the appropriate dividers.

Could It be a hardware issue that need to be addressed by Asus or is it a problem with Sandra reporting incorrect values ?



Fok Speling Misstake

 

[Quetzacoatl]

hmm, i'm not familiar with that board, do you have to set the PCI/AGP divider in the bios, or is it automatic?

Soon enough, Intel will make the i845s...imagine dual channel Sdram...*shudder*

 

[phsstpok]

Maybe Sandra doesn't recognize the KT333 chipset and just misreports the information.

<b>I type sixty words per minute. Ten are spelled correctly.</b>

 

[Crashman]

I'm betting that you need to make MANUAL changes in BIOS to get the lower PCI divider of 1/5.

<font color=blue>You're posting in a forum with class. It may be third class, but it's still class!</font color=blue>

 

[SidVicious]

In the available FSB list, a PCI value is assigned to each FSB, so I don't have access to any PCI divider. I guess the BIOS automaticaly apply the corresponding divider based on the selected FSB speed. I hate it when I end up having no control on some of my BIOS setting...



Fok Speling Misstake

 

[Crashman]

You do have an adjustment, you just didn't know it. Under the CPU/PCI Frequency menu you get more than one option for 166MHz. Heck, if you just key down to that menu and hit the enter key, it should give you a menu of available speeds.

<font color=blue>You're posting in a forum with class. It may be third class, but it's still class!</font color=blue>

 

[SidVicious]

I already knew that menu, it's the FSB/PCI speed list I was talking about in my last post. According to that menu the 1/5 PCI divider is supposed to kick in at 150/30 and up, logically, choosing 166/33 should give me 33/66 PCI/AGP but for some reason I end up with 42/84...

I'll try to boot with a 150MHz FSB than i'll drop you guys a line...

P.S. I re-flashed the BIOS to 1014 and cleared the CMOS just in case, but that did'nt solved my problem.

Fok Speling Misstake

 

[SidVicious]

I'm now running on 150/30 according to the BIOS but Sandra reports 150/38. I'm beginning to wonder if the 1/5 multiplier actually exist on the A7V333-R. That may mean that Asus will be in deep [peep] when users will beggin plugging 166MHz FSB T-Bred Bs on their A7V333...

I'm lucky, i'm not experiencing any bugs or crashes from running my Supra 56k modem and Visiontek GeForce2 MX400 on OCed busses.



Fok Speling Misstake

 

[SidVicious]

UPDATE

No luck yet, tried different FSB settings over 133MHz, all the way up to 175MHz ( where my 128Mb PC2100 Samsung memory decided it was too high ) and none seem to "trigger" the 1/5 divider, I guess my hardware will have to live with a screaming fast 42MHz PCI bus.

Here's a mission for our lurking link hunters out there =) Sandra could be reporting false values, that's why I would need another app to double check Sandra's report of the PCI/AGP speed.

I'm sure I'm not the only one running an A7V333 on a 166MHz FSB, I would appreciate any reports of fellow users running that board at the same speed.



Fok Speling Misstake

 

[tombance]

I would have thought most cards would be freezing/crashing by then. Sandra probably doesnt incorporate the 333 chipset and there recgonises anything over 133 as an oc and therefore does the calculations for the PCI/AGP buses accordingly.

My sig's faster than yours, and it overclocks better too....

 

[SidVicious]

That's what I tought too but we can't put the blame on Sandra without some hard facts or double checking to back these suspicions.

That's why I asked fellow users to post links to similar programs that can report PCI/AGP bus frequencies and to report their own values when running the A7V333 on a 166MHz FSB.


Fok Speling Misstake

 

[phsstpok]

Looks like an old answer in the SiSoft FAQ but it might still apply.

"Q: Sandra reports incorrect PCI bus speeds.
A: Check that FSB is correct. Certain boards run the PCI bus asynchronously and Sandra may be unable to detect this in all cases. If you're using a board which allows you to play with the PCI clock (e.g. 1/3 & 1/4 settings) the detection may not work - depending on clock generator chip. Update to the latest version of Sandra."



<b>I type sixty words per minute. Ten are spelled correctly.</b>

 

[SidVicious]

Woah, sorry about making such a big deal about that, I neglected to carefully read through Sandra's FAQ. I usually read such litterature with great care, I must have skipped this section =/

I would need recommendations for some kind of test program that will report the real PCI/AGP speeds, I just wanna make sure everything is all right.

Anyway, Ty all for showing support, it's greatly appreciated =)

Fok Speling Misstake

 

[phsstpok]

No biggy, I remembered when Sandra had trouble when KT133A was introduced so I just thought something similar was happening.

I don't know of any programs which I am sure read true speeds. Many just seem to read BIOS settings and video driver settings. As the Sandra notes mention the program probably needs to detect the capabilities of the clock generator in order to produce correct results. (I wonder if boards which can lock AGP and PCI have a separate clock generator for these)?

This is going to sound silly but when I wanted to be sure AGP was where I thought it was. I would use an old DOS based 2D benchmark <b>[EDIT - I mean video memory benchmark. It isn't graphical]</b>. Since there is no software acceleration then the benchmark depends on bandwidth, minus video card latencies. If the benchmark number goes up then I knew AGP was running faster. Crude but it let me know when the dividers were changing. I assumed the PCI divider changed at the same time as AGP. On my KT133A board the dividers changed from 2/3 and 1/3, AGP and PCI respectively, to 1/2 and 1/4 at 128 Mhz (I think it was 128 Mhz) and not 133. This explained why I could run FSB at 128 Mhz but not 120-127 Mhz. Even though I had the CPU at the same level of overclock.

The reason I was playing with FSB below 133 Mhz was I found I could run a Tbird 1.0 at slightly over 1.7 Ghz (for a very short time) at 128 Mhz but couldn't go over about 1.68 Ghz with any other FSB speed. (I didn't try every combination; something like 130 Mhz, 133 Mhz, 140 Mhz, 145 Mhz, 150, Mhz, and 154 Mhz were the other ones I tried besides 128 Mhz and lower).

At the time the version of Sandra I had did not detect the correct AGP/PCI/ISA speeds. Come to think of it I don't know if it does now.

<b>I type sixty words per minute. Ten are spelled correctly.</b><P ID="edit"><FONT SIZE=-1><EM>Edited by phsstpok on 10/26/02 03:23 PM.</EM></FONT></P>

 

[CMRvet]

I have MSI Kt3 Ultra at 166 MHz and SANDRA is also reporting the PCI at 42 MHz and AGP at 84 MHz.
However I'm sure that this mobo has the 1/5 divisor.

For me it is a SANDRA problem.


<b>(<font color=yellow>as good as it looks</font color=yellow>)</b>