Submit Hint Search The Forums LinksStatsPollsHeadlinesRSS
14,000 hints and counting!

Click here to return to the 'Best way to cool your iMac' hint
The following comments are owned by whoever posted them. This site is not responsible for what they say.
Best way to cool your iMac
Authored by: hexeric on Jun 07, '11 08:05:27AM

i find this much better: (settings here also work for boot camp/win7!)

and yes: unfortunately, an imac (just as an macbook) definitely needs cooling. the HD and GPU fans are as good as not adjusted by heat sensors. i do a lot of video work on my new imac (quad core) and keeping the minimum fan speed of both, HD and GPU (it's controlled by the DVD drive fan interestingly) at 2000rpm is sufficient. CPU is monitored by the application and the fan adjusts automatically.

further: my old iMac (v2007,1) got black stripes on the display due to the heat issue - you know, when the diffusion skin in front of the LEDs gets a little burnt...

[ Reply to This | # ]
Best way to cool your iMac
Authored by: bluescreen on Jun 19, '11 11:05:32AM

In rasterized video display generator systems stationary 'stripes' (like boxes) are generally indicative of either memory array (draw buffer[s]) or output array write, read or output addressing-counter failures. In such 2 dimensional flat 'table' arrays, each row and column location (and blocks of the same) is/are formed of data applied at numerical x-y addresses.

Each numerical 'x' or 'y' (row or column) range of addresses represent some vertical or horizontal 'stripe' range, as any specific 'block' represents an intersection of the coincidence of those two ranges of addressing-counter regions. The lowest numbered physical address (start) is in one corner and the highest numbered address (end) is at the diagonally opposite corner of the stored or drawn 'x-y table array'.

In a (more simply described for example) simplified base-10 example, addressing-counter numbers are generally formed/generated by some sort of a serial to parallel decoding array that reuses the 'fine' 0-10 or 0-100 (hottest, fastest, hardest-working 'fine' address-counting) counter by merely occasionally incrementally adding a 'coarse' 10-100, 100-1,000, 1,000-10,000, 10,000-100,000 (etc) 'bit' to the address-count output to get it up to the higher address-count location-range numbers. Repeating 'stripes' are thus indicative of a (repetitive) failure of the underlying 'fine' address-counting numbering decoder.

If the 'fine' (0-100 ex) decoder starts missing it's ability to generate addressing numbers, say for example, from '50-100' during each count-up sequence, then the array will be striped (empty of data) in equal stripes all the way up the range, since those array locations cannot be addressed to be read, rewritten or output.

Stripes don't indicate a failure of the display, they indicate a failure of the display's own controller/driver circuits, the GPU or the graphics card's output or the display's input/output interface.

Otherwise the failure to refresh, redraw, erase or move a moving or movable block of data like a program window or animation graphic element within it (a so-called 'sprite-block' of local image data) or deal with it's overlay-depth-priority or transparency (box-trails) is a problem in the RAM addressing, rewriting or data manipulation/flow handling of the GPU (or data or software instruction) itself.

Since modern low single voltage DRAMs don't heat up like their older progenitors 90% of these sorts of (non-defect) failures are voltage or thermal parallel addressing counter/connection related.

Edited on Jun 19, '11 11:11:59AM by bluescreen

[ Reply to This | # ]