System: Dual Xeon Irwingdale 3.4 GHz, Chipset E7525, 4GB Ram
OS: Windows XP Pro SP2
Fact Rows: 20,000,000
Aggregations: 1300
Processing Time: 3 Hrs
System: Dual AMD Opteron 246 (2.0 GHz), AMD Chipset 8000 (MB Tyan S2885). 4 GB RAM
OS: Windows 2003 Enterprise (SP1)
Fact Rows: 20,000,000 (same data set as 1st case)
Aggregations: 1300
Processing Time: 6-7 Hrs
Both test environments have the:
Same RAM speed
Same model HDDs
Is this difference due to any lack of AMD-specific optimizations in SSAS binaries? or some other reason?
Also is there any way to reduce the thread priority on SSAS binary during optimization? It seems to almost completely hose the machine (windows 2003 ent.) during processing. When attempted to change processing priority through task manager I got access denier error (logged in as admin)Did you test under the same OS and 100% equial another options?
Such difference as you have acounted seems to be too screwy.|||
Vladimir Chtepa wrote:
Did you test under the same OS and 100% equial another options? Such difference as you have acounted seems to be too screwy.
The operating systems are different - Windows XP SP2 vs. Windows 2003 SP1.
SSAS options are all same.
Also when its processing on Windows 2003 SP1/AMD Opteron test environment it is much harder to use any other applications (open windows, task manager, etc) than on the Windows XP. The AMD/Windows2003 gets very hosed/hung up when processing at 100% CPU.
On other hand the IntelXeon/Win XP case it still responsive even when processing faster and at 100% CPU for hours.
No comments:
Post a Comment