I have two machines running milkyway@home the first being an 95 2500k @4.8ghz and a GTX 580, the second being an i7 3930k @ 5ghz and a GTX 690. What doesn;t make sense to me however, is how the first machine is seemingly doing better than the second.

http://milkyway.cs.rpi.edu/milkyway/hosts_user.php?userid=446602

3930k details--http://milkyway.cs.rpi.edu/milkyway/host_app_versions.php?hostid=463265

2500k details--http://milkyway.cs.rpi.edu/milkyway/host_app_versions.php?hostid=463149

As far as I can see, the discrepancy could be due to the fact that the i5 has 242 valid tasks, the i7 only 227.
The turn-around time of the i5 is also lower (0.11 days) than the one for the i7 (0.29 days). More throughput is more credit is more recent average credit.

---

Is this fixable?

I have two machines running milkyway@home the first being an 95 2500k @4.8ghz and a GTX 580, the second being an i7 3930k @ 5ghz and a GTX 690. What doesn;t make sense to me however, is how the first machine is seemingly doing better than the second.

http://milkyway.cs.rpi.edu/milkyway/hosts_user.php?userid=446602

3930k details--http://milkyway.cs.rpi.edu/milkyway/host_app_versions.php?hostid=463265

2500k details--http://milkyway.cs.rpi.edu/milkyway/host_app_versions.php?hostid=463149

Hey. Milkyway@home project requires a Double prediction support from graphics card. Both graphics cards (580 & 690) support DP calculation, but the gtx 580 is designed for DP calculation better than 690. If you want more out from 690 gpu use it for single prediction projects. Example: gpugrid.net project. gpugrid.net do not need DP, so that project works really well on gtx 690.

Hey. Milkyway@home project requires a Double prediction support from graphics card. Both graphics cards (580 & 690) support DP calculation, but the gtx 580 is designed for DP calculation better than 690. If you want more out from 690 gpu use it for single prediction projects. Example: gpugrid.net project. gpugrid.net do not need DP, so that project works really well on gtx 690.

The word would be double/single precision, but with that correction, the explanation makes sense. To get the fastest DP processing on an NVidia card, you need to purchase from the Tesla range, not GeForce.