SI
SI
discoversearch

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor.  We ask that you disable ad blocking while on Silicon Investor in the best interests of our community.  If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.
Politics : Formerly About Advanced Micro Devices -- Ignore unavailable to you. Want to Upgrade?


To: Tenchusatsu who wrote (73262)9/27/1999 6:17:00 PM
From: Petz  Respond to of 1578881
 
Ten, re:<Are you claiming that a 33% overclocked AGP 2x bus is faster than a non-overclocked AGP 4x?>

Ten: In certain cases, yes

Petz: <Most BX boards when running at 133 MHz DO NOT OVERCLOCK THE AGP or even the PCI slots. They use a 1/4 divider for the PCI and the AGP runs at twice that speed, instead of the normal 1/3 divider.

So your entire reasoning is faulty.>

Ten: Uh, no. 440BX does not have a 1/2 divider for AGP (or a 1/4 divider for PCI). All 440BX boards that run at 133 MHz must overclock AGP and PCI.

Sorry, here's a quote from Anand's review of the Freetech P6F107 BX board:
But for those of you willing to try, settings above 124 are available with a 1/4 PCI ratio to keep things as close to spec as possible. That PCI speed is reported right next to the FSB speed in the BIOS. All settings are available with any CPU installed, but don't forget to set the AGP divider via the aforementioned jumper...

Also, on bxboards.com's comparison of five BX boards for overclocking:
At 133Mhz, all boards offered a PCI clock divider of /4 by default, or offered a choice of speed. A PCI /4 divider is very desirable for future proofing, as this keeps the PCI bus running at the speed it was intended - 33Mhz. At 150Mhz [FSB], most AGP cards will not function, even if an AGP clock of 2/3 is used.

Reading between the lines, you were right about the AGP bus which is being overclocked 33%, but not about the PCI bus and the IDE interface. So the fact that 33 partially overclocked BX comes very close to i820 performance, which supposedly has twice the bandwidth of BX is proof that for most applications bandwidth doesn't matter as much as latency.

Petz