I've been looking through some of my Pending Results.

http://setiathome.berkeley.edu/workunit.php?wuid=129642593 is pending with 2 sucessful results returned. Validate state. Checked, but no consensus yet.

I was a bit surprised to see that both were processed by Chicken 2.2B:

http://setiathome.berkeley.edu/result.php?resultid=535660949
http://setiathome.berkeley.edu/result.php?resultid=535660950

I have seen this before, but not where both clients were using the same app.

---

Sir Arthur C Clarke 1917-2008

It doesn't happen a lot but is not unheard of, even when all the hosts on the WU are running stock.

Alinator

I've been looking through some of my Pending Results.

http://setiathome.berkeley.edu/workunit.php?wuid=129642593 is pending with 2 sucessful results returned. Validate state. Checked, but no consensus yet.

I was a bit surprised to see that both were processed by Chicken 2.2B:

http://setiathome.berkeley.edu/result.php?resultid=535660949
http://setiathome.berkeley.edu/result.php?resultid=535660950

I have seen this before, but not where both clients were using the same app.

Revision: 2.2B|xT+GFX|FFT:IPP_SSE3|Ben-Joe
CPUID: Intel Core 2 Duo 'Conroe 2M'

Revision: 2.2B|xK+GFX|FFT:IPP_SSE|Ben-Joe
CPUID: AMD Athlon XP 'Thoroughbred'

They're not really the same app, The xT build used by the Conroe would not run on the Athlon. Even if both were using the SSE version the Athlon would only access up to SSE versions of optimized routines while the Conroe would be using SSE2 and SSE3 versions.

The host which will probably resolve the issue is running stock 5.15 and seems likely to finish any minute now. I won't be surprised if all 3 get credit, a very small difference can block consensus.

                                                              Joe

Revision: 2.2B|xT+GFX|FFT:IPP_SSE3|Ben-Joe
CPUID: Intel Core 2 Duo 'Conroe 2M'

Revision: 2.2B|xK+GFX|FFT:IPP_SSE|Ben-Joe
CPUID: AMD Athlon XP 'Thoroughbred'

They're not really the same app, The xT build used by the Conroe would not run on the Athlon. Even if both were using the SSE version the Athlon would only access up to SSE versions of optimized routines while the Conroe would be using SSE2 and SSE3 versions.

The host which will probably resolve the issue is running stock 5.15 and seems likely to finish any minute now. I won't be surprised if all 3 get credit, a very small difference can block consensus.

                                                              Joe

Thanks Joe, I was hoping that you or Simon would reply to this thread.

I am a great fan of your optimized apps, I was just a little surprised to see this particular result. If I understand it correctly the current state of this WU means that the results are weakly similar i.e. not invalid but not valid either. When the third result arrives the other results will get checked again.

I have read about this happening when different OS or app version were involved.

I see now that the other host is MS Vista Home, mine is XP Pro.
Also mine is AMD, other is Intel.
Mine has 1 core, other has 2 cores.
Mine only has SSE, other has SSE, SSE2, SSE3, SSSE3

I will be interested to see the result when the third result arrives.

Keith.

The third result is in. All 3 results are Valid. Interesting to see that it is the third result that has been chosen as the canonical result.

---

Sir Arthur C Clarke 1917-2008

I'm going to speculate because it was the second pass through validation, the new arrival was used as the 'base' result for comparision.

This seems to be a change from the way the validator worked when it was on Solaris (I'm assuming Ptolemy is running Linux). Previously the lowest RID was usually selected as the canonical when a WU would go through a 'nominal' validation sequence. This implies the RID was used as the base result, and now it seems the newest arrival is used.

In any event, it doesn't seem to make a difference in the percentage of canonicals my hosts get over time.

Alinator