Changes between Version 17 and Version 18 of GpuWorkFetch

Timestamp:

Dec 29, 2008, 10:53:34 AM (15 years ago)

Author:

davea

Comment:

--

GpuWorkFetch

@@ -26 +26 @@
 This document proposes a modification to the work-fetch system that solves these problems.
 
-For simplicity, the design assumes that there is only one GPU type (CUDA).
-It is straightforward to extend the design to handle additional GPU types.
+For simplicity, the design considers only one GPU type (CUDA).
+However, it is straightforward to extend the design to handle additional GPU types.
 
 == Terminology ==
@@ -62 +62 @@
 New abstraction: '''processing resource''' or PRSC.
 There are two processing resource types: CPU and CUDA.
+
+The notion of long-term debt
 
 === Per-resource-type backoff ===
@@ -79 +81 @@
 This is stored in an object of class PRSC_WORK_FETCH.
 
-Data members of PRSC_WORK_FETCH
+Data members of PRSC_WORK_FETCH:
 
 '''ninstances'''
@@ -122 +124 @@
 It has the following "persistent" members (i.e., saved in state file):
 
-'''double long_term_debt*'''
-
 '''backoff timer'''*:  how long to wait until ask project for work specifically for this PRSC;
 double this any time we ask for work for this rsc and get none
@@ -143 +143 @@
 }}}
 
+Each project has the following work-fetch-related state:
+
+'''double long_term_debt*''': the amount of processing (including GPU, but expressed in terms of CPU seconds) owed to this project.
+
 === debt accounting ===
 {{{
-for each resource type
-   R.accumulate_debt(dt)
+for each resource type R
+   for each project P
+      if P is not backed off for R
+         P.R.LTD += share
+   for each running job J, project P
+      for each resource R used by J
+         P.R.LTD -= share*dt
 }}}
 
@@ -209 +218 @@
    clear backoff for the PRSC of each returned job
 }}}
+
 == Scheduler changes ==
 {{{