>>>>Now my build times have gone back to 24-25 seconds instead of 8-9 secs. Wierd but I can live with that. Remember that my new HW is still unpacked an hour's drive away.
>>>
>>>Now that sounds like HD trouble - either trying to read from a nearly dead sector or trying to find "clean" sectors to write on.
>>
>>Why does it sound like HW trouble? I can hear the disk when its working. It's not.
>>
>
>Assuming the compile was the only major load in your old machine (no video or music en/decoding running at the same time, no virus sending spam all over the place, no defrag running at compile time) perf of cpu/memory/video is nearly constant. Only with the HD you can have wildly fluctuating results, as HD perf is influenced by
>
>the amount of free space left,
>the pattern and localization (absolute and relative) of free space left,
>the pattern and localization (absolute and relative) of the files needed to read,
>the fragmetation of the files needed to read
>(the pattern and localization (absolute and relative) of a dynamic swap disk can influence memory speed as well)
>
>IF there are a few "brittle" sectors on the HD the OS will try to read/write again and again, and this is especially expensive, as the operation breaks the read ahead pattern of the HW cache of the disk.
>
>There are also few special cases if the OS is installed on a small separate partition or disk, as sometimes the "personal" folders of the OS have to fight for space on an otherwise nearly empty data disk. Hint: dynamic swap disks are often on that partition.
>
>Easy eplanations could be that you deleted a few GB of old, unneccesary data in preparation for the HW move to the new machine or cleaned ot your temp files and internet caches <g>.
>
>If you want to reuse the old HDs and or plan to have the old pc still working, I'ld try to get to the bottom of those compiling speed differences - such fluctuations are mostly explainable or can be a symptom of bad pc health. And data resurrection can be very costly.
>
>regards
>
>thomas

>>>>Now my build times have gone back to 24-25 seconds instead of 8-9 secs. Wierd but I can live with that. Remember that my new HW is still unpacked an hour's drive away.
>>>
>>>Now that sounds like HD trouble - either trying to read from a nearly dead sector or trying to find "clean" sectors to write on.
>>
>>Why does it sound like HW trouble? I can hear the disk when its working. It's not.
>>
>
>Assuming the compile was the only major load in your old machine (no video or music en/decoding running at the same time, no virus sending spam all over the place, no defrag running at compile time) perf of cpu/memory/video is nearly constant. Only with the HD you can have wildly fluctuating results, as HD perf is influenced by
>
>the amount of free space left,
>the pattern and localization (absolute and relative) of free space left,
>the pattern and localization (absolute and relative) of the files needed to read,
>the fragmetation of the files needed to read
>(the pattern and localization (absolute and relative) of a dynamic swap disk can influence memory speed as well)
>
>IF there are a few "brittle" sectors on the HD the OS will try to read/write again and again, and this is especially expensive, as the operation breaks the read ahead pattern of the HW cache of the disk.
>
>There are also few special cases if the OS is installed on a small separate partition or disk, as sometimes the "personal" folders of the OS have to fight for space on an otherwise nearly empty data disk. Hint: dynamic swap disks are often on that partition.
>
>Easy eplanations could be that you deleted a few GB of old, unneccesary data in preparation for the HW move to the new machine or cleaned ot your temp files and internet caches <g>.
>
>If you want to reuse the old HDs and or plan to have the old pc still working, I'ld try to get to the bottom of those compiling speed differences - such fluctuations are mostly explainable or can be a symptom of bad pc health. And data resurrection can be very costly.
>
>regards
>
>thomas

Just an update on an old conversation. I got my new Dell machine in (dual-core sweet) some time back and observed the same fluctuation in build time except that on the new machine it would go from 2.3 seconds to 4.5 seconds and back. That pretty much rules out the hardware.

I haven't seen the higher time in quite a while now. Considering that a few months ago my build time was 160 seconds or more and now it is about two seconds, you can imagine that I am one happy camper.

BTW, there were two things that reduced my build time considerably (besides the new hardware).

One, using only two include (.H) files instead of a dozen of so.

Two, organizing the PRG classes into about 30 files instead of 150 or so.

Peter