Buy him a decent bottle of Brandy not Grand Marnier

Asbach should be on the list. If I was going to benefit I would think about it, but sorry JD.

---

Heck, if this port of the code works, we should all chip in Buy him a bottle :)

And put the cork back in while there was still a sip or two left when we give it to him eh?

Thanks for offering to save me a sip... quite gentlemanly of you. If serious, drink of choice at home is a fine MACALLAN Single Malt, the older the better and
just one sip is enough ;-)

Seriously though, We've all got Alex Kan (AK) and others to thank, I just worked persistantly for one day to get the src-v8 tarball to build in the Windows environment. Nothing magic about it. Something is still amiss, and I can't quite put my finger on it. Now I'm having to try to understand portions of the source, and I'm far from being a mathematician. It is much easier to "port" existing code than it is to develop it firsthand, or understand it. Hat's off to the guys & gals that understand this stuff.

If anyone is aware of bugs or gotchas in the code pointed to above, please point me to a fix and I'll try to make this happen. I think that I've resolved the convolution/correlation issue in gaussFit.cpp. Someone correct me if I'm wrong, the operator is symmetric so correlation becomes convolution , either operation will work. The only issue is to allocate enough memory for the output and to reposition to exclude taper zones. (if using ippsConv())

thanks again for the support...

The original E4500 host chip has moved from the ECS MoBO. Yep the UPS guy delivered the replacement for the failed Gigabyte GA-P35-DS3L. I tried to keep the clock the same @2420MHz with 2x1GB DDR2-800 Patriot. Switch was made at 21 Mar 23:00 UTC. For those keeping close tabs, you might notice a shift in the numbers at that time. Possibly due to the higher memory speed or different chipset.

@Pilot - I think I got some of those "HOT" WUs you referred to in another thread and took this MoBo down. Timeframe matches ~Mar 9... Nevermind that the Q6600 was running 3600MHz on air with the memory screaming at 1200MHz. CoreTemp was good 61-56C on PRIME and passed STABLE ...then along comes a HOT wu... and like Pat Travers sings "Out go the Lights".

Cheers all,
JDWhale

---

On msattler's Penryn host, some recent returned results show:

Windows optimized S@H Enhanced application by Alex Kan
Version info: OS X SSSE3 (Intel, Core 2-optimized v8-nographics) V5.13 by Alex Kan

Sadly, almost all are in a very, very narrow Angle Range range near .39, but at that range they are taking about 78% as much CPU time as what Mark was running before. About 2820 CPU seconds vs. about 3620.

Nice, nice.

Is this an Alex Kan port to Windows? JDWhale port of Alex Kan's Mac code? someone else?

This version does not seem to report clock rate to stderr out, so I'm assuming Mark has not changed from his recent 4590 in making these comments. If he has suddenly greatly upped his clock rate (ummm... not likely, I think) then the ap improvement may be less.

This improvement at this AR should not be taken to predict the overall average improvement across AR mix, though this AR range is quite important to overall performance.

---

Arch, this is JD's port of Alex's V8 app.

---

Is JD's port of Alex's V8 code available for others to trial?

I have a couple of Quads ... a Penryn and a older QX6,700 extreme.

I assume Mark is trialing the ported Mac to Windows code on his Phase cooled Quaddie (Penryn) only?

---

It's good to be back amongst friends and colleagues

Curmudge, I'm not sure if JD has gone to Vegas yet, but I think he is still tweaking with the code.

---

Nice, nice.

Is this an Alex Kan port to Windows? JDWhale port of Alex Kan's Mac code? someone else?

I wish I could claim credit, but I've not provided anyone with what I've got so far. Currently I'm only crunching some of the low and high angle ranges, whereas these are mid-angle. There appears to be more than one port being tested. This is a good thing! [edit]I've got a feeling that Mark is testing some penryn optimized (ssse4) code, but I could be and often am wrong.[/edit]

I'm leaving for Las Vegas Sunday, still loading up the farm with work for while I'm gone.

JDWhale

---

On msattler's Penryn host, some recent returned results show:

Windows optimized S@H Enhanced application by Alex Kan
Version info: OS X SSSE3 (Intel, Core 2-optimized v8-nographics) V5.13 by Alex Kan

Sadly, almost all are in a very, very narrow Angle Range range near .39, but at that range they are taking about 78% as much CPU time as what Mark was running before. About 2820 CPU seconds vs. about 3620.

Nice, nice.

Is this an Alex Kan port to Windows? JDWhale port of Alex Kan's Mac code? someone else?

This version does not seem to report clock rate to stderr out, so I'm assuming Mark has not changed from his recent 4590 in making these comments. If he has suddenly greatly upped his clock rate (ummm... not likely, I think) then the ap improvement may be less.

This improvement at this AR should not be taken to predict the overall average improvement across AR mix, though this AR range is quite important to overall performance.

The kitties waved their magic wand and turned my PC into a Mac...........LOL.

Actually, the app I am testing is a beta port of Alex's code to Windows done by our own jason gee..........he has been working on it for a while now over on the lunatics site. It is not complete yet nor ready to release, but the preliminary testing I did looked good, and he gave me permission to test it live for a bit while he is still working on it. All results seem to be validating properly, although it is overclaiming credit by a small amount.

And no, the clock rate has not changed, still running at 4.58ghz...

---

"Freedom is just Chaos, with better lighting." Alan Dean Foster

Congratulations Jason... Looking good my friend!!!


@Mark - Your Seti City Kitty is now on Starbucks payroll ?

A clear improvement I see from JG's app. working on the premise of the app, at standard clock rate of 3.0GHZ your looking at around 5350 seconds for 72.6 dredits, a big improvement.

Congrates all you guys trying to match us Mac guys. You'll catch us soon enough i fear.

---

A clear improvement I see from JGs app. working on the premise of the app, at standard clock rate of 3.0GHZ your looking at around 5350 seconds for 72.6 credits, a big improvement.

Congrats all you guys trying to match us Mac guys. You'll catch us soon enough i fear.

Yeah, It's looking like It might be possible, Afterall the Colonial Vipers are programmable for more speed along with their maneuverability I'd think. ;) :)

---

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

Congratulations Jason... Looking good my friend!!!


@Mark - Your Seti City Kitty is now on Starbucks payroll ?

LOL.......you probably don't visit the Cafe too often, but in my kitty thread there I announced the rescue of another fine kitty from the local shelter......he is named Starbucks......my GF Lori adopted him, and we brought him home yesterday.
And the banner kitties thought they would say hello to him......

And so far, Jason's code is looking really fast.......waiting to see if it hits some other AR WUs.......I am not at home right now, so I don't know what's coming up in the cache for the next few hours...

---

"Freedom is just Chaos, with better lighting." Alan Dean Foster

Let's see here...........

In the beginning was only the stock Boinc/Seti app. No optimized app was available anywhere in all the land. All was well and good. Then some wonderful people went to work and created an optimized app. All was still well and good, only much faster. Then the Berkeley staff said "Wait a minute! We are giving away too many useless credits and this cannot be tolerated." So the wise men reduced the number of credits awarded across the entire project to compensate for the faster app that everyone was using.

Now the possibility exists of a Windows app, created from the fastest known source code on the planet? A Windows app again created by some wonderful people? The Berkeley staff is going to have to slow us all down again!!

Let's see here...........

In the beginning was only the stock Boinc/Seti app. No optimized app was available anywhere in all the land. All was well and good. Then some wonderful people went to work and created an optimized app. All was still well and good, only much faster. Then the Berkeley staff said "Wait a minute! We are giving away too many useless credits and this cannot be tolerated." So the wise men reduced the number of credits awarded across the entire project to compensate for the faster app that everyone was using.

Now the possibility exists of a Windows app, created from the fastest known source code on the planet? A Windows app again created by some wonderful people? The Berkeley staff is going to have to slow us all down again!!

Hopefully this can slip under their radar, In Stealth Mode as It were. :)

---

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

Heck, if this port of the code works, we should all chip in Buy him a bottle :)

And put the cork back in while there was still a sip or two left when we give it to him eh?

Thanks for offering to save me a sip... quite gentlemanly of you. If serious, drink of choice at home is a fine MACALLAN Single Malt, the older the better and
just one sip is enough ;-)


@Pilot - I think I got some of those "HOT" WUs you referred to in another thread and took this MoBo down. Timeframe matches ~Mar 9... Nevermind that the Q6600 was running 3600MHz on air with the memory screaming at 1200MHz. CoreTemp was good 61-56C on PRIME and passed STABLE ...then along comes a HOT wu... and like Pat Travers sings "Out go the Lights".

Cheers all,
JDWhale

Something took out my main MB a P5K shortly after that post. The CPU, and Memory and other stuff still ok and running on a spare P5B at a moderate OC at the moment, or at least till I get my P5K back from ASUS.
I have tested the Power Supply and other stuff, and they are all running well with the same cooling.
Can't really say if the High temp readings were the cause or result of the failed MB.

---

When we finally figure it all out, all the rules will change and we can start all over again.

Curmudge, I'm not sure if JD has gone to Vegas yet, but I think he is still tweaking with the code.

Thanks for the post Satan, and congrats on breaching the 10K RAC. I will have to wait until I can afford the 16 way Nelham when I plan to get the beast in the Spring or Summer 09.

In the meantime I hope JD has a nice break from Sunday, and I look forwards to both JD's and JG's Alex code port being released.

Unlike Mark, my Penryn is air cooled, but I have had it running stably at 4.0GHz for a couple of months.

---

It's good to be back amongst friends and colleagues

Pilot wrote:

Can't really say if the High temp readings were the cause or result of the failed MB.

I prefer to think that Berkeley let some "Fire Breathing" WUs get out. LOL!

And so far, Jason's code is looking really fast.......waiting to see if it hits some other AR WUs.......I am not at home right now, so I don't know what's coming up in the cache for the next few hours...

Any reason you switched back away from it? Recent returns look like this:

<stderr_txt>
Optimized SETI@Home Enhanced application
Optimizers: Ben Herndon, Josef Segur, Alex Kan, Simon Zadra
   Version: Windows SSE4 64-bit based on S@H V5.15  'Noo? No - Ni!'
  Revision: R-2.4V|xS|FFT:IPP_SSE4|Ben-Joe
     CPUID: Intel(R) Core(TM)2 Extreme CPU X9650  @ 3.00GHz 
     Speed: 4 x 4581 MHz 
     Cache: L1=64K L2=6144K
  Features: MMX SSE SSE2 SSE3 x86_64 

---

And so far, Jason's code is looking really fast.......waiting to see if it hits some other AR WUs.......I am not at home right now, so I don't know what's coming up in the cache for the next few hours...

Any reason you switched back away from it? Recent returns look like this:

<stderr_txt>
Optimized SETI@Home Enhanced application
Optimizers: Ben Herndon, Josef Segur, Alex Kan, Simon Zadra
   Version: Windows SSE4 64-bit based on S@H V5.15  'Noo? No - Ni!'
  Revision: R-2.4V|xS|FFT:IPP_SSE4|Ben-Joe
     CPUID: Intel(R) Core(TM)2 Extreme CPU X9650  @ 3.00GHz 
     Speed: 4 x 4581 MHz 
     Cache: L1=64K L2=6144K
  Features: MMX SSE SSE2 SSE3 x86_64 

Because it was intended to be a beta test only, and not quite ready for prime time..........(not that it did not appear to be working very well, aside from a small credit overclaiming issue)........so I have gone back for now to the Crunch3r app I had been running for months.
Stay tuned for further developements...........

---

"Freedom is just Chaos, with better lighting." Alan Dean Foster

And so far, Jason's code is looking really fast.......waiting to see if it hits some other AR WUs.......I am not at home right now, so I don't know what's coming up in the cache for the next few hours...

Any reason you switched back away from it? Recent returns look like this:

<stderr_txt>
Optimized SETI@Home Enhanced application
Optimizers: Ben Herndon, Josef Segur, Alex Kan, Simon Zadra
   Version: Windows SSE4 64-bit based on S@H V5.15  'Noo? No - Ni!'
  Revision: R-2.4V|xS|FFT:IPP_SSE4|Ben-Joe
     CPUID: Intel(R) Core(TM)2 Extreme CPU X9650  @ 3.00GHz 
     Speed: 4 x 4581 MHz 
     Cache: L1=64K L2=6144K
  Features: MMX SSE SSE2 SSE3 x86_64 

Because it was intended to be a beta test only, and not quite ready for prime time..........(not that it did not appear to be working very well, aside from a small credit overclaiming issue)........so I have gone back for now to the Crunch3r app I had been running for months.
Stay tuned for further developments...........

Yeah, I think we'll stay tuned to this Kitty Channel.

---

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