I've swapped an A64 3000 (Venice) for an A64 X2 3800.
The 3000 was overclocked to 2.25GHz. It ran in the low to mid 30s idle and ~40 under load.
The 3800 is at stock.
I used MX-1 paste as according to the instructions for it and was getting temps of 36-7 idle and 52 under load. I used Everest and another monitoring program, as well as S&M's own monitoring, just to be sure. Slight variances, but no more than 1C. Speedfan reported another temperature, labelled "core", which went up as high as 62 and jumped 14C as soon as S&M started.
I removed the HSF, cleaned it and the IHS, then reapplied using much less MX-1 because the instructions called for a lot of it and I was entertaining the idea that the instructions were for a smaller tube. They said that 8% of the tube is used, so if they were for a smaller tube then I'd used too much. It certainly looked like too much. So I used about half.
Idle is now 40-41 and 55 under load. That other temp from Speedfan goes up as high as 65.
I'm at a loss to explain it. Nothing else has changed since I had the 3000 in and the 3800 at stock shouldn't be that much hotter than a 3000 with a 25% overclock.
Any ideas?
EDIT: More data. I ran Orthos small FFTs as a stress test which would allow me to vary the speed of the CPU fan (S&M, for some reason, stopped if I set my CPU fan over 1600rpm).
Varying the fan speed from 1100 to 2000 had no effect on the "temp2" reading, which I (and all the hardware monitoring programs I used) read as a CPU temp. It did, however, drop the "core" reading by 3C. I really hope that isn't a genuine reading - up to 65C at stock?
The dip in the red line near the middle is while the CPU fan was running at 2000rpm instead of 1100rpm.
The 3000 was overclocked to 2.25GHz. It ran in the low to mid 30s idle and ~40 under load.
The 3800 is at stock.
I used MX-1 paste as according to the instructions for it and was getting temps of 36-7 idle and 52 under load. I used Everest and another monitoring program, as well as S&M's own monitoring, just to be sure. Slight variances, but no more than 1C. Speedfan reported another temperature, labelled "core", which went up as high as 62 and jumped 14C as soon as S&M started.
I removed the HSF, cleaned it and the IHS, then reapplied using much less MX-1 because the instructions called for a lot of it and I was entertaining the idea that the instructions were for a smaller tube. They said that 8% of the tube is used, so if they were for a smaller tube then I'd used too much. It certainly looked like too much. So I used about half.
Idle is now 40-41 and 55 under load. That other temp from Speedfan goes up as high as 65.
I'm at a loss to explain it. Nothing else has changed since I had the 3000 in and the 3800 at stock shouldn't be that much hotter than a 3000 with a 25% overclock.
Any ideas?
EDIT: More data. I ran Orthos small FFTs as a stress test which would allow me to vary the speed of the CPU fan (S&M, for some reason, stopped if I set my CPU fan over 1600rpm).
Varying the fan speed from 1100 to 2000 had no effect on the "temp2" reading, which I (and all the hardware monitoring programs I used) read as a CPU temp. It did, however, drop the "core" reading by 3C. I really hope that isn't a genuine reading - up to 65C at stock?
The dip in the red line near the middle is while the CPU fan was running at 2000rpm instead of 1100rpm.
Last edited: