Still NO Invalid results for S41.06 after about 100 results :)
Since you started all this Akos, I have only had one invalid result back on 13apr06 using D40. Does not seem that long ago we where trying to beat the 4x mark. Now we are at the 6x mark. WoW.
Standard app @ 280 min / S41.06 @ 44 min = 6.3636 x faster
:)
S41.06 is the best yet for older single-core CPUs, and I haven't seen any errors. Look at what it did for this old 700 MHz P-III -- shaved about an hour off the already outstanding performance of S40.12 on long WUs!!! A 3.6 GHz Xeon with the standard client took 2.6 times as long as my 700 to crunch this WU! But hyperthreaded chips still do best on S40.04 or S-39L.
But hyperthreaded chips still do best on S40.04 or S-39L.
Yes, but why should the HT be used? At least for Prescott 3.4 GHz - if you turn the HT off, you will be significantly faster with S40.07 than with HT_enabled_S39L.
P4, 3.0 Mhz (1 CPU of
)
P4, 3.0 Mhz (1 CPU of HT)
first 3 results S41.07 faster S41.06 about 3.4 % (0.966)
S41.06/07
Valid - 10
Invalid - 0
Good work!
First true S41.07 result,
)
First true S41.07 result, defo slower than S41.06 on my PIII 801 mhz
From 11168 (06) to 11570 (07)
Will stick with 06 on this pc
AMD64 3000 + Newcastle core
)
AMD64 3000 + Newcastle core 10% overclock
S41.07
Fastest result ever and validated. No comparison with S41.06 cos that did only z1 wu's
http://einsteinathome.org/task/27884402
2214 sec. avg over 6 results.
Results comparing S41.07 with
)
Results comparing S41.07 with S41.06:
Athlon64 3400 - S41.07 is 5% faster
Athlon64 2800 - S41.07 is 6% faster
Sempron 2600 90nm - S41.07 is 6% faster
Sempron 2500 (older) - S41.07 is 0% faster
Athlon 2500 (Barton) - S41.07 is 0% faster
Interesting, the newer cores are all 5-6% faster with S41.07
The older cores are the same with both clients
One much older machine running a Tbird core:
D41.14 is running 6.5% faster than D41.13
All work units validating with no errors so far on the above versions.
Thanks Akos for all the hard work & great clients!
Regards/Beyond
Opteron 275 U41.04
)
Opteron 275
U41.04 2130-2210
S41.06 2250
D41.13 3000
D41.12 2600
S40.12 3100 ~seconds
Athlon 64 3800+
S41.07 1900
S41.06 2200
D41.13 2600
D41.12 2400
S40.12 2850 ~seconds
Lots of validated results no invalid for S41.07 or U41.04
Still NO Invalid results for
)
Still NO Invalid results for S41.06 after about 100 results :)
Since you started all this Akos, I have only had one invalid result back on 13apr06 using D40. Does not seem that long ago we where trying to beat the 4x mark. Now we are at the 6x mark. WoW.
Standard app @ 280 min / S41.06 @ 44 min = 6.3636 x faster
:)
98SE XP2500+ @ 2.1 GHz Boinc v5.8.8
S41.06 is the best yet for
)
S41.06 is the best yet for older single-core CPUs, and I haven't seen any errors. Look at what it did for this old 700 MHz P-III -- shaved about an hour off the already outstanding performance of S40.12 on long WUs!!! A 3.6 GHz Xeon with the standard client took 2.6 times as long as my 700 to crunch this WU! But hyperthreaded chips still do best on S40.04 or S-39L.
RE: But hyperthreaded chips
)
Yes, but why should the HT be used? At least for Prescott 3.4 GHz - if you turn the HT off, you will be significantly faster with S40.07 than with HT_enabled_S39L.
Could somebody try out this
)
Could somebody try out this version of S41.07 on a HT enabled system?
RE: Could somebody try out
)
sure,
partial should be done baking in about 5 seconds.
other unit of the HT seems to be cooking along nicely.
http://einsteinathome.org/task/28292140