The SCSI controllers take away load from the CPU as they can take over the bus without bugging the CPU with this lowly transport stuff ;-)
I have a 2940UW somewhere and a few Atlas IV 18GB, maybe some day I'll try that myself. It's still deep in some boxes from moving though, so it might take some time (I'm not a good packer - well, I'm not from WI ;-) )
My Barton 2800+ (running under Millenium) is faster
wu r1_1213 :
with S39L -> 4207 sec to 4900 sec
with S39L + S41.06 -> first wu -> 3,892.96 sec (1h04min52sec) ; wu valid and credits
with S41.06 -> second wu r1_1213.0 -> 2,970.00 sec !!! (0h49min30sec)
My Barton 2800+ shows similar - S40.12 was around 61 minutes / WU, first S41.06 was 48 minutes (not yet validated).
The SCSI controllers take away load from the CPU as they can take over the bus without bugging the CPU with this lowly transport stuff ;-)
I have a 2940UW somewhere and a few Atlas IV 18GB, maybe some day I'll try that myself. It's still deep in some boxes from moving though, so it might take some time (I'm not a good packer - well, I'm not from WI ;-) )
You can get the same effect with standard hardrives and a good ($100+) ide pci/pcie harddrive controller. If you're doing standardPCI keep in mind that the PIC bus is only as wide as 2 7200rpm drives as sustained read/write speed and you loose all the burst benefits from a larger cache/
First result with S41.06 (mixed - S40.12): 1427 sec. vs ave 1600 sec 40.12 only. Result is valid. Will monitor further results.
AMD XP1600+
Second result (these are short WUs) S41.06 only: 1293
Excellent speedup!
Wow! First LONG WU (from r1_1325.0...) ran 4141 sec. with S41.06 vs previous WU from the same dataset using S40.12 taking 5283 sec. Almost 22% improvement!!!
Dual Xeon HT 2200/400 (Foster
)
Dual Xeon HT 2200/400 (Foster core, S603), PC800 Rambus (RIMM-RAM)
~115 Min. with S41.06
~138 Min. with S40.12
~157 Min. with S40.04
~172 Min. with S39L
All values for the long results.
Quite fast compared to the Prestonia, I guess the RAM makes a difference.
I heard that SCSI speeds up those Prestonias a lot too.
RE: I heard that SCSI
)
What would SCSI have to do with the Einstein@home crunching speed?
MrS
Scanning for our furry friends since Jan 2002
The SCSI controllers take
)
The SCSI controllers take away load from the CPU as they can take over the bus without bugging the CPU with this lowly transport stuff ;-)
I have a 2940UW somewhere and a few Atlas IV 18GB, maybe some day I'll try that myself. It's still deep in some boxes from moving though, so it might take some time (I'm not a good packer - well, I'm not from WI ;-) )
Ananas, could you post the
)
Ananas, could you post the CPU-Z cache info for those Fosters?
RE: My Barton 2800+
)
My Barton 2800+ shows similar - S40.12 was around 61 minutes / WU, first S41.06 was 48 minutes (not yet validated).
RE: The SCSI controllers
)
You can get the same effect with standard hardrives and a good ($100+) ide pci/pcie harddrive controller. If you're doing standardPCI keep in mind that the PIC bus is only as wide as 2 7200rpm drives as sustained read/write speed and you loose all the burst benefits from a larger cache/
@Mr.Pernod: I was wrong,
)
@Mr.Pernod:
I was wrong, CpuZ says it's a Prestonia too - I always thought Prestonia starts with S604 - sorry for the confusion.
CpuZ report
CpuZ says "S604" but it's an S603 IWILL DP400 - at least that I'm sure about *g
First result with S41.06
)
First result with S41.06 (mixed - S40.12): 1427 sec. vs ave 1600 sec 40.12 only. Result is valid. Will monitor further results.
AMD XP1600+
Seti Classic Final Total: 11446 WU.
RE: First result with
)
Second result (these are short WUs) S41.06 only: 1293
Excellent speedup!
Seti Classic Final Total: 11446 WU.
RE: RE: First result with
)
Wow! First LONG WU (from r1_1325.0...) ran 4141 sec. with S41.06 vs previous WU from the same dataset using S40.12 taking 5283 sec. Almost 22% improvement!!!
Seti Classic Final Total: 11446 WU.