So, I was working on a P10469 (HD7970)project and noticed that my TPF was inconsistent. I saw that there was an occasional gpu usage
dropout in MSI afterburner. I thought it was a bad overclock, but noticed that it happened at stock clocks. I changed the F@h cpu
priority to below average from low a well as reserving and it . Frame times were a more consistent
6:40 instead of bouncing between 7:00 and 6:40. The downside is that my temps went up about 10*C!
Sorry if this was common knowledge. i really don't understand why the cpu spikes so high on occasion.
Also, looked at the p10469 times in the overclock.net databaseand graphed them in excel to see what frame times are needed to get
200k ppd on that project. You need 5:59, which would melt my poor hd 7970
The trendline formula is sloppy on excel. It should be 1.32 billion * t^-1.495 = y
y is ppd and t is tpf.
to calculte t for a certain milestone (for example 400,000 ppd):
t= -1.495 (sqrt) of y/x or 400,000/1.32 billion in this example. t would be 226 seconds or 3:46
dropout in MSI afterburner. I thought it was a bad overclock, but noticed that it happened at stock clocks. I changed the F@h cpu
priority to below average from low a well as reserving and it . Frame times were a more consistent
6:40 instead of bouncing between 7:00 and 6:40. The downside is that my temps went up about 10*C!
Sorry if this was common knowledge. i really don't understand why the cpu spikes so high on occasion.
Also, looked at the p10469 times in the overclock.net databaseand graphed them in excel to see what frame times are needed to get
200k ppd on that project. You need 5:59, which would melt my poor hd 7970
The trendline formula is sloppy on excel. It should be 1.32 billion * t^-1.495 = y
y is ppd and t is tpf.
to calculte t for a certain milestone (for example 400,000 ppd):
t= -1.495 (sqrt) of y/x or 400,000/1.32 billion in this example. t would be 226 seconds or 3:46
