Francois may well download the new app and take number one pot, but we'll know who did the real work.

Fred, he is still running the original 2.4 code, the times of that machine are slower than my Mac Pro.

I think sadly, despite all the hype, there is nothing to show.

But 20% faster than the same machine was when running 2.4V back in October!! Hence my suggestion that it is NOT running the 2.4V now.

F.

---

V the T.V series was crap and V for Vendetta was even crapper, so things are looking up already.

Francois, IF you finally produce the goods then many people including myself will owe you an apology.

---

Ahhh...so V is the man formally known as who? (aka Dr_who?). Welcome back Francois. I hope we see your SKT kick some mac A.... with either your code or the Whale port of Alex's code.

Oh dear! Has our tester-in-chief been naughty and got himself banished again?!!

Well here are the first results I have pulled off the Q6600 using the latest pre-release App:


Direct Link

Looks like this is confirming my initial impressions from the Penny that this iteration is slightly slower that its predecessor. Another run on the Penny under way but I may not get round to posting until after my zzzzzzz's.

F.

---

And don't forget who? still has the fastest WINTEL machine around.....

i ll only open my mouth when i am done with my own version.

my new name says it all.

V

Welcome...

Guy Fawkes was born April 13, was that the day you assumed the new name in his honor? Then I can only assume your "final version" will be on November 5, as a final tribute.

Good luck,
JDWhale

I just have some family from Corsica, so, I know the real meaning of Vendetta.
I am mostly a viking descendant, from Normandy, This explains my barbarian manners ... :-P I am from the village of William the conquerer, I grow up next to the castle of "Robert le Diable", Duke of Normandy

V


Francois

Ah yes, William, That's My middle name, Turns out I've got roots in 3 countries, Ireland, England and yep France(My last name is Bobier and It's supposed to be French), I can trace My paternal ancestry back to 1770 Ireland(Wexford) and from what I've read the family was there from the 12th century and in France as early as the 10th, Before that the trail goes cold, On the Maternal side I only know a few things like My grandmothers maiden name and that both maternal families had family crests(Coan(Moms maiden name) and Spafford(Grandmas maiden name)) and are reputed to have been English nobility of some sort at one time and one of My Grandmothers ancestors came to Texas in a Conestoga wagon across the prairie to get to Texas, Before eventually arriving in California.

In any case I'm glad to hear that Your Seti app is doing better than 2.4V, Not that 2.4V is all that bad, But improvement is always welcome none the less.

---

The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's

Ah well the "vac" did its work on the Penny while I was catching up on the threads.

Here is what I have so far, compared to the last (what I take to be) "equivalent" results:


Direct Link.

Note that in the mid-AR's, all the results for the second column fall between the max and min results for the first column and any given (mid) AR band so I would say there is no significant difference.

The earlier results are ones that I captured around 13 April.

F.

---

Oh dear! Has our tester-in-chief been naughty and got himself banished again?!!

Looks very likely that you are correct!

---

It's good to be back amongst friends and colleagues

Oh dear! Has our tester-in-chief been naughty and got himself banished again?!!

Looks very likely that you are correct!

Yeah no sign of him, or his posts?,once again. Well if you need a couple of extra Q6600's for testing while he's away you know where to find me :)

---

he (who?) was barely ahead of me in the boincstats page and then he disappeared..
i was wondering how fast his RAC went up to do that..


oh well
another spot i will pass.

---

he (who?) was barely ahead of me in the boincstats page and then he disappeared..
i was wondering how fast his RAC went up to do that..


oh well
another spot i will pass.

He (Who?) is now at number 8 in the top_hosts and now identified as V.

F.

---

A message of comfort to those running SSE3 but who can't run SSSE3 (note this may not apply if you have AMD and cannot circumvent the "naughty Intel").

In a post here JDWhale agreed with my estimate that his host Ginger (P4-540 (Prescott H/T) @ 3200MHz) should see an improvement of about 30% at mid-AR's from running the new code. Well he is currently running the latest pre-release of super-fried Chicken on both Ginger and Skipper (P4D-820 @ 2800MHz) and the early results indicate we neither of us realized what was coming.

Bearing in mind that there are very few results in yet, the average improvement for both these hosts on mid-AR WU's has been around 67%! Both are currently crunching 5 mid-AR WU's in much less time that it took for 3 with the best currently available SSE3 Op App.

Of course, I will post pictures when more data (and, hopefully, more AR's) are available.

F.

---

A message of comfort to those running SSE3 but who can't run SSSE3 (note this may not apply if you have AMD and cannot circumvent the "naughty Intel").
...

Just to clarify, for sse3-capable AMD users it appears this won't be a problem. The SSE3 build is non-Intel-specific , no patching needed.

Comparisons of the generic-sse3 versus patched-p4-specific-SSE3 builds turned out identical in performance. So much so that even when the Intel-only build was patched to run on AMD-Phenom as an experiment, that it still ran identically (well) to the generic-sse3. [same result by comparison of same builds on a PD].

So we took the Intel-only-sse3 build out the back and shot it. It has since been purged from the source code configurations.

I'm not entirely clear on the history of the 'Naughty Intel' issue, but it seems it is not there now. If you choose the proper compiler settings, which do not seem to compromise performance by any measure I can tell, then the same build is suitable on p4(sse3 capable) or AMD(sse3 capable) [Though Intel don't guarantee this].

Jason

---

"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.

Just to clarify, for sse3-capable AMD users it appears this won't be a problem. The SSE3 build is non-Intel-specific , no patching needed.

Comparisons of the generic-sse3 versus patched-p4-specific-SSE3 builds turned out identical in performance. So much so that even when the Intel-only build was patched to run on AMD-Phenom as an experiment, that it still ran identically (well) to the generic-sse3. [same result by comparison of same builds on a PD].

So we took the Intel-only-sse3 build out the back and shot it. It has since been purged from the source code configurations.

I'm not entirely clear on the history of the 'Naughty Intel' issue, but it seems it is not there now. If you choose the proper compiler settings, which do not seem to compromise performance by any measure I can tell, then the same build is suitable on p4(sse3 capable) or AMD(sse3 capable) [Though Intel don't guarantee this].
Jason

Awwww! Jason.

No more entertaining posts from ML1 on this issue?

Thanks for clarifying that one. Now a lot more crunchers will be even more grateful to you guys.

F.

---

A message of comfort to those running SSE3 but who can't run SSSE3 ...

<snip>

Bearing in mind that there are very few results in yet, the average improvement for both these hosts on mid-AR WU's has been around 67%! Both are currently crunching 5 mid-AR WU's in much less time that it took for 3 with the best currently available SSE3 Op App.

Of course, I will post pictures when more data (and, hopefully, more AR's) are available.

F.

Can't slip anything by my ol' buddy Fred W. [...mumble...grumble...]

While running one of my convoluted benchmark ideas, I switched to running the latest SSE3 release candidate inside BOINC (single process only) on both Ginger and Skipper for a few hours as your detective work points out... I'd switched back to running AK_WhalePort V0.2S after the benchmarks run on both hosts as further detective work will confirm.

Since being discovered, I'll now switch back to running the new SSE3 release candidate for a good portion of the day, this time on both processes live. This should help stabilize the numbers a bit, though I doubt there is much variation in WU ARs in the queue.

I hadn't looked in detail at the "numbers" you've quoted. "I couldn't see the forest for the trees", so to speak.... Details...Details...Details.

If you wouldn't mind, please keep an eye on Lovey also. Preliminary benchmarks on the E4500 chip are mixed, so I'll be running 3 different release candidates (SSE3, SSSE3, SSSE3X) here... your pictures will tell a better story than I can gather only by looking at the numbers.

Happy Sunday/Monday as applies,
JDWhale

---

Just to clarify, for sse3-capable AMD users it appears this won't be a problem. The SSE3 build is non-Intel-specific , no patching needed.

Comparisons of the generic-sse3 versus patched-p4-specific-SSE3 builds turned out identical in performance. So much so that even when the Intel-only build was patched to run on AMD-Phenom as an experiment, that it still ran identically (well) to the generic-sse3. [same result by comparison of same builds on a PD].

Now that is very interesting...

So we took the Intel-only-sse3 build out the back and shot it. It has since been purged from the source code configurations.

So you might say that it has bit the dust?

I'm not entirely clear on the history of the 'Naughty Intel' issue, but it seems it is not there now. If you choose the proper compiler settings, which do not seem to compromise performance by any measure I can tell, then the same build is suitable on p4(sse3 capable) or AMD(sse3 capable) [Though Intel don't guarantee this].

That's one little "Naughty Intel" that ran for far far too long. It wouldn't surprise me if a "class action" suit was to follow for recompence against all the resources needlessly wasted previously...

Awwww! Jason.

No more entertaining posts from ML1 on this issue?

Will the "Bit the dust" do? :-)

Thanks for clarifying that one. Now a lot more crunchers will be even more grateful to you guys.

Very much so, agreed!


As for the 67% speedup?!... Wow, if that one settles out to be really the case, then good food indeed!

Happy crunchin',
Martin

---

See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)

<snip>
If you wouldn't mind, please keep an eye on Lovey also. Preliminary benchmarks on the E4500 chip are mixed, so I'll be running 3 different release candidates (SSE3, SSSE3, SSSE3X) here... your pictures will tell a better story than I can gather only by looking at the numbers.

Happy Sunday/Monday as applies,
JDWhale

Yes, I noticed it was running on Lovey as well, but there were only 2 results in when I did my run and previously Lovey seemed to have been running "Xeon" builds so I didn't think there was valid direct comparison?

I will keep an eye on it.

F.

---

<snip>
If you wouldn't mind, please keep an eye on Lovey also. Preliminary benchmarks on the E4500 chip are mixed, so I'll be running 3 different release candidates (SSE3, SSSE3, SSSE3X) here... your pictures will tell a better story than I can gather only by looking at the numbers.

Happy Sunday/Monday as applies,
JDWhale

Yes, I noticed it was running on Lovey as well, but there were only 2 results in when I did my run and previously Lovey seemed to have been running "Xeon" builds so I didn't think there was valid direct comparison?

I will keep an eye on it.

F.

The "Xeon" clients actually only refer to an additional AK optimization that at the time might have been intended to steer clear of Core2 with the limited L2 cache, etc. Intel has released so many variations of Core2, the optimization might now have a name other than "Xeon".

Yes, The AK_WhalePort Xeon is/was faster than the Core2 build... After building SSE3 for Ginger & Skipper, I never thought to try SSE3 on Lovey. Until I ran the benchmarks on the new Release Candidates (RC) last night I didn't think that SSE3 would benefit the E4500 Core2 Duo.

My numbers are mixed, thus I'll be running each of three new RCs on Lovey.. My guess is the smaller L2 cache might benefit at some ARs with SSE3. I thought others might benefit from my experiments when it comes to choosing the "best" client for some of the more economical Core2 Duos.

As Murphy would have it... I was going to post my findings/results at Lunatics first thing this morning, but I've been exposed. (Jason might not let me "sample the cookie dough" next time :-( )

I'm considering a similar test for Wrongway, but you didn't hear me say that. Thurston is out of the question, he is staying 100% AK_WhalePort in his quest to the top 20 (then there might be a change).

Cheers
JDWhale

---

..."sample the cookie dough"...

beware the SSE3 RC1 'generic' has some eXtra herbs also, and is not quite equivalent to previous SSE3 builds.

---

"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.

How about a weekly update with just minimal babble...

With the switch to RC1 SSE3 estimated RAC should be updated once again for Skipper, Lovey, and Wrongway; but I haven't plugged the data into my unofficial guestimator yet. We should be seeing some updated plots/charts if/when the data fills in the AR bins.

Host = Skipper: P4D-820 @ 2800MHz


06Apr My estimated RAC = 1280
13Apr My revised estimated RAC = 1342



Host = Lovey: E4500 @ 2420MHz


06Apr My estimated RAC = 2300
13Apr My revised estimated RAC = 2720



Host = Wrongway: Q6600 @ 2520MHz


06Apr My estimated RAC = 4700
13Apr My revised estimated RAC = 5040

Note here... As I'm writing, Wrongway is still running benchmarks for the Lunatics crew. After review of said benchmarks, I'll switch BOINC to run an RC1 client of my choosing (probably the one with the best performance in the mid-AR WUs we're seeing so many of.

Warning : if there is an onslaught of VHAR or VLAR WU's I won't hesitate to switch to the best client for the data. Same holds true for Lovey.



Host = Thurston: Q6600 @ 3375MHz


06Apr My estimated RAC = ~7400
13Apr My revised estimated RAC = 7080

Remember - Random unattended "Reset project" ~40 hours ago :-(
At least it's going in the right direction!

However tempting to switch... Thurston will remain on AK_WhalePort V0.2X until official Lunatics release... I want to see how accurate my estimated RAC of 7080 is.



Host = Ginger: P4-540 (Prescott H/T) @ 3200MHz

Sorry, Ginger is in the powder room and doesn't want her picture took.

Trust me, her graph is climbing, but since she's been switching between KWSN R-2.4 (SSE3) and AK_WhalePort V0.2S (SSE3), and now running RC1 (SSE3), it's not a reflection of a single clients performance. Trust the charts/graphs posted by the experts.


Regards,
JDWhale

---

I'll add you to my monitoring list and post updates when there is anything to report, if that is OK by you?

I am curious to see whether the "Xeon effect" is ubiquitous.

F.

@Fred--Thanks & OK by me. I'll let you know when I switch over to AK's Xeon v8.

QS

@Fred
Switched over to Alex's Xeon v8 at 21 Apr 2008 4:06:45 UTC. Anything returned after that time stamp will have been crunched with the Xeon flavor.

BTW, I show 268 WUs with valid results (other than -9s) returned in the last 3.5 days. Two hundred of those WUs had angle ranges of 0.36 through 0.45.

QS

---