http://setiathome.berkeley.edu/workunit.php?wuid=144812504
I understand that a Conroe is far better CPU than Presler- and that ChickenApp will give you major computational boost- but that is just incredible!
Unfortunately the result that surpassed even that has been deleted already, but I once had 88.70 cobblestones result with difference of 33k vs. 3.7k seconds CPU time- is that actually possible?

Rambo

Yes, it is possible. On that workunit, the 'slow' host is using what looks like an older, 'stock' app (v5.12, I am guessing its the current 64-bit linux app).
The 'fast' one is using an optimized app:

Version: Linux 64-bit based on S@H V5.15 'Noo? No - Ni!'
Revision: R-2.2B|xT|FFT:IPP_SSSE3|Ben-Joe

I am not sure what the stock app uses, but since it is compiled to be 'one size fits all', I would guess that it isn't using any variant of SSE.

The 'Chicken App' is using SSSE3 AND Intel's IPP libraries (MUCH MUCH faster than GCC's equivalents).

Add that to the much better CPU on the 'fast' one, and frankly I am surprised that it isn't even more of a speed blowout.

---

https://youtu.be/iY57ErBkFFE

#Texit

Don't blame me, I voted for Johnson(L) in 2016.

Truth is dangerous... especially when it challenges those in power.

That is about right see resultid=587149712 from my C2D.

Andy

That is about right see resultid=587149712 from my C2D.

Andy

Both of you need to update your Chicken. You are using 2.0, but your 'opponent' is still using 1.3.... 2.4 is out. :)

---

https://youtu.be/iY57ErBkFFE

#Texit

Don't blame me, I voted for Johnson(L) in 2016.

Truth is dangerous... especially when it challenges those in power.

Ah-HA! So time to switch :), Would love to squeeze some more juice from the old D :D

---

That is about right see resultid=587149712 from my C2D.

Andy

Both of you need to update your Chicken. You are using 2.0, but your 'opponent' is still using 1.3.... 2.4 is out. :)

Thats an older unit from my account I wanted one at same AR. Been using chicken noodle since about 0400 UTC.

It reduces crunch time on Pent M and C2D to about 73% of stock.

Andy

http://setiathome.berkeley.edu/workunit.php?wuid=144812504
I understand that a Conroe is far better CPU than Presler- and that ChickenApp will give you major computational boost- but that is just incredible!
Unfortunately the result that surpassed even that has been deleted already, but I once had 88.70 cobblestones result with difference of 33k vs. 3.7k seconds CPU time- is that actually possible?

Rambo

Sure the stock app was fairly much slower than the chicken optimised apps, and AK's MAc versions too. This difference is said to be going to be a bit less with the current releases. Most of all the core2 cpus rock & overclock well too, so your 10 to 1 improvement seems quite practical, but figures would vary depending on the system. Also consider you'd then be running on 2 cores simultaneously effectively doubling output( or four if you chose to get a quad).

So if your Optimised App + Core2 Cpu + OC gives you [hypothetical] 10 times the output per core that's 20(for a core2duo, OR 40 TIMES for a core2quadTOTAL OUTPUT .

[ Divide those by 2 against presler with 2 cores, I was was thinking of my single core slugs :D ]

---

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.

ChickenApp 2.4 unfortunately dies painfully on my machine while 2.2-SSE2 kills the machine (unusually heavy load), so building my own and see what it will bring :)

---

Update to this: as it turns out, with machine-optimized client (DIY) the "real" crunch time of my client is comparable with Conroe now(!)- BELOW 4k secs-1.5h for 56 cobblestones unit), however... somehow the time for completion is not CPU time, but time from sending workunit to receiving result(!).

Needless to say, I'm now a little lost- or how can decrease intervals of reporting results (connecting every 0.05 day is pretty much often)?

---