🚀 go-pugleaf

T&thee (smiteq@gil.ipswichcity.qld.gov.au) wrote:

> The DMA clock speed is altered by increasing the crystal frequency. This may 
> have no effect as some DMA clocks run slow from new but... Norton Utilities 
> does not provide evaluation of the DMA clock reports. Over running the DMA 
> clock WILL result in premature failure... The board will fry itself.

Usually the DMA clock runs from a divided down CPU clock, or from the ISA
clock. If the motherboard/BIOS is able to handle 33MHz it would select a
different divisor and everthing should be fine. A good check is if the
motherboard correctly identifies the CPU speed at boot time.

> The second problem is the increased heat generated by a CPU running 25% faster 
> than it's rated speed.

A fan is always a good idea. Especially with Australian temperatures.

> As I understand the Chip selection process, slow CPU chips have failed factory 
> testing at higher speeds... This is to say all chips have a manufactured 
> "speed target". Those that fail say: a 33 Mhz set of tests may see life as a 
> 25 Mhz chip and those 33s with a faulty FPU might perform as a 25Mhz SX.
> Given that the subject crystal was 50 Mhz, the CPU must have been 25 Mhz... A 
> chip that the makers considered unsuitable to sell as a 33.

There is a slight wrinkle to this. In the case of the Pentium 90/100, demand
for the 90 is so much higher than the 100. This means that a lot of parts
that could have ended up in the 100 bin, end up in the 90 bin to make up
for the demand. However those parts are most likely only tested to 90 MHz.
-- 
--      .     +------------------------------------------------------------+
    ,--_|\    | John Saunders  - (Work) johns@rd.scitec.com.au             |
   /  Oz  \   |                - (Home) john@pacer.apana.org.au            |
   \_,--\_/   | SCITEC LIMITED - Phone +61 2 428 9563 - Fax +61 2 428 9933 |
         v    +------------------------------------------------------------+