FSB reporting on Intel 430TX - how?

[trodas]

On Asus TXP4-X mainboard, witch use Intel 430TX chipset, latest CPU-Z does report the FSB wrongly. A good example is my P90 running at 75x1.5 - reported as 55x2 = 110MHz: http://valid.x86.fr/ia4gwh

 

So I "look around" to see, what others use and on the similar, Asus TXP4 mainboard held Antimony first place with 83MHz:

http://hwbot.org/submission/2373822_

 

The 83MHz settings looks familiar, my board can do 7.14, 50, 55, 60, 66, 75 and 83MHz, so except my envy that the IDE controller made it on his board, I have nothing against that score. I looked that the FSB is correctly reported using CPU-Z v1.63 - so I grab that version on CPU-Z page (last one in history) and run it too:

 

...but it does not show anything! What the hell?! :mad:

 

That does not make slightest sense. Yes, the v1.63 somewhat ignored the L2 cache on Antimony's Asus TXP4, but that it is. It is weird, in the place where cache IO should be ( http://s18.postimg.org/4yf1lv18p/Asus_TXP4.jpg ) - between the CPU socket and battery, he have a different IO on the photo: http://img.hwbot.org/u8730/image_id_949660.jpeg

No idea what is going on there?

 

And also the CPU-Z does not report the L2 at all for him, even that the board should come with it, as mentioned on page 8 of the manual:

"Level 2 Cache: 512KB Pipelined Burst SRAM onboard."

http://www.mediafire.com/?1sds42nqtnn7jpq

 

Another score (CPU clock) is also weird on the Asus TXP4:

http://hwbot.org/submission/2221547_patriot219_cpu_frequency_k6_2_500mhz_499.83_mhz

Correct FSB, no ram info and weird L2 of side 128kBy?! (1.58):

 

And with same CPU-Z used it suddently report no FSB and no L2 info (besides almost standard no ram info):

http://hwbot.org/submission/2221534_patriot219_wprime___32m_pentium_1_100mhz_26min_18sec_383ms

 

 

 

Another very similar board is Asus TX97-XE:

 

No FSB and L2 in CPU clock (1.46):

http://hwbot.org/submission/754366_azesmbog_cpu_frequency_k6_233mhz_%28model_6%29_265.6_mhz

 

No FSB but correct L2 in PiFast (1.46):

http://hwbot.org/submission/754370_azesmbog_pifast_k6_233mhz_%28model_6%29_23min_20sec_980ms

 

No FSB but correct L2 in SPi 1M (1.46):

http://hwbot.org/submission/754386_azesmbog_superpi___1m_k6_233mhz_%28model_6%29_10min_30sec_430ms

 

No FSB but correct L2 in SPi 32M (1.46):

http://hwbot.org/submission/753943_azesmbog_superpi___32m_pentium_1_150mhz_12h_5min_54sec_720ms

 

No FSB and no ram info, but correct L2 in wPrime 1024M (1.46):

http://hwbot.org/submission/757107_azesmbog_wprime___1024m_k6_233mhz_%28model_6%29_13h_32min_53sec_352ms

 

...

 

 

While IMHO none of the scores seems suspect (all are within the possibility of these boards), what seems odd is the variously different results from FSB detection and some cache oddity.

What I would love to know is, what CPU-Z should I use to get the FSB reported properly, as sometimes works for some with various old CPU-Z versions, while I get ZERO with the old versions of CPU-Z

 

Anyone know?

[trodas]

CPU/FSB detection using CPU-Z versions 1.61.4 - 1.63 - 1.66 - 1.70 - 1.72.1 - 1.73:

 

 

In short, since 1.70 it start the detection of wrong clock (110 instead of 112.5 as 75x1.5 gives), yet added detection of FSB, witch is wrong, when overclocked and right when running on stock (last image and validation):

http://valid.x86.fr/5hhxk9

 

...that bring back the question on how that can work for someone before and how to do it now?

[Strunkenbold]

Theres no special magic behind this. FSB was always shown on Socket 7 boards but just in combination with K6-2 and up cpus. All others showed no FSB. But still right frequency. L2 cache is shown with K6-2 but disappeared with K6-3+ and vice versa.

This is just another example how hard it is for Frank to support these old boards. But what me make start thinking is that totally wrong results are being validated.

And as always the only thing todo is report bugs to Frank or in the CPU-Z Bugs thread.