walaka7 said:I did all but three frames of that one on the dothan and it was a dedicated run, no playing around with goodies. I wanted convergance to have little effect on the overall picture as the frame times are considerably shorter. And i concur on the bandwidth limitations as it appears to be the issue. Im not sure how much more testing im gonna do on the dothan and QMD's as you can't just get them on their own, they have to be piped or in my case already there with the proc swap. Ram timing was not the tightest.. 2-3-3-6 but the testing was done on a 533 version of the dothan and memory was @ 200 mhz
Oh and yes SSE2 was enabled
I really meant steps not frames. QMDs all have 100 frames but the number of total steps and the steps per frame vary. The time per step is consistent after convergence (actually it's pretty consistent preconvergence but about 25% faster than postconvergence steps). All QMDs start with 2000 steps but the total number caclulated prior to convergence. Some wind up with close to 2000 steps, most wind up with about 2070 steps, but some wind up with nearly 2200 steps. On a 2070 step WU 10 frames will have 207 steps meaning 7 of those frames have 21 steps and 3 will have 20 steps. The 20 step frames will take 5% less time to complete than the 21 step frames. A 2200 step WU would converge in frame 9 and take about 10% longer than a 2000 step WU. PPD would be about 10% less. That's why I asked how many steps, to see if you had an average, short or long QMD.
I wondered if SSE2 would be enabled since the client misidentifies the Dothan as a P3 but you have answered my question.
Here's a sample from the log.
Code:
A 2068 step QMD
[03:18:23] - Number of total steps will change until convergence
[03:19:28] Completed 0 out of 2000 steps (0)
[03:35:13] Completed 21 out of 2021 steps (1)
[03:35:13] Writing local files
[03:50:16] Completed 41 out of 2041 steps (2)
[03:50:16] Writing local files
[04:06:12] Completed 62 out of 2062 steps (3)
[04:06:12] Writing local files
[04:10:48] Timered checkpoint triggered.
[04:10:52] WF converged, jumping to MD
[04:10:52] Verifying checksum
[04:10:52] Finished
[04:11:53] Completed 68 out of 2068 steps (3)
[04:26:49] Completed 83 out of 2068 steps (4)
[04:26:49] Writing local files
[04:47:46] Completed 104 out of 2068 steps (5)
[04:47:46] Writing local files
[05:08:42] Completed 125 out of 2068 steps (6)
[05:08:42] Writing local files
[05:28:39] Completed 145 out of 2068 steps (7)
A 2192 step QMD
[16:58:53] - Number of total steps will change until convergence
[16:59:40] Completed 0 out of 2000 steps (0)
[17:10:19] Completed 21 out of 2021 steps (1)
[17:10:19] Writing local files
[17:20:32] Completed 41 out of 2041 steps (2)
[17:20:32] Writing local files
[17:31:10] Completed 62 out of 2062 steps (3)
[17:31:10] Writing local files
[17:42:25] Completed 84 out of 2084 steps (4)
[17:42:25] Writing local files
[17:53:33] Completed 106 out of 2106 steps (5)
[17:53:33] Writing local files
[17:59:53] - Autosending finished units...
[17:59:53] Trying to send all finished work units
[17:59:53] + No unsent completed units remaining.
[17:59:53] - Autosend completed
[18:04:50] Completed 128 out of 2128 steps (6)
[18:04:50] Writing local files
[18:16:33] Completed 151 out of 2151 steps (7)
[18:16:33] Writing local files
[18:28:15] Completed 174 out of 2174 steps (8)
[18:28:15] Writing local files
[18:37:21] Timered checkpoint triggered.
[18:37:24] WF converged, jumping to MD
[18:37:24] Verifying checksum
[18:37:24] Finished
[18:38:06] Completed 192 out of 2192 steps (8)
[18:42:08] Completed 198 out of 2192 steps (9)
[18:42:08] Writing local files
[18:56:57] Completed 220 out of 2192 steps (10)
[18:56:57] Writing local files
[19:11:45] Completed 242 out of 2192 steps (11)
[19:11:45] Writing local files
[19:26:33] Completed 264 out of 2192 steps (12)
[19:26:33] Writing local files
Last edited: