Hi Brock,
I didn't update anything and the old units were half done completed as usual. all the rest in my que have been going no problem.

@hiamp-

Seems like you made a more graceful upgrade than I did. I think my problems were mostly related to not having the right app_info.xml file in place. I'll remember that for next time. Could be that there was something else going on too, like whatever is still bothering the Batman.

---

to hiamps, lurkers and all that can crunch without errors :)

please post your experience comparing GenSSE2, IntelSSE2 patched and IntelSSE3, my teammates (several SETI+AMD crunchers) would find it very useful. Thank you!

---

to hiamps, lurkers and all that can crunch without errors :)

please post your experience comparing GenSSE2, IntelSSE2 patched and IntelSSE3, my teammates (several SETI+AMD crunchers) would find it very useful. Thank you!

I will run the SSE3 for a few more days then try the SSE2 Intel and let you know.

---

Official Abuser of Boinc Buttons...
And no good credit hound!

@ SargeD > running chicken SSE2 w/ BOINC v 5.8.11

Private Data 2/12/2007 3:20:30 PM||Starting BOINC client version 5.8.11 for windows_intelx86
2/12/2007 3:20:30 PM||log flags: task, file_xfer, sched_ops
2/12/2007 3:20:30 PM||Libraries: libcurl/7.16.0 OpenSSL/0.9.8a zlib/1.2.3
2/12/2007 3:20:30 PM||Data directory: C:\\Program Files\\BOINC
2/12/2007 3:20:31 PM|SETI@home|Found app_info.xml; using anonymous platform
2/12/2007 3:20:35 PM||Processor: 1 GenuineIntel Intel(R) Pentium(R) 4 CPU 2.40GHz
[x86 Family 15 Model 4 Stepping 1] [fpu tsc sse sse2 mmx]
2/12/2007 3:20:35 PM||Memory: 1023.48 MB physical, 5.40 GB virtual
2/12/2007 3:20:35 PM||Disk: 128.00 GB total, 118.45 GB free
2/12/2007 3:20:35 PM|SETI@home|URL: http://setiathome.berkeley.edu/; Computer ID: 3083566; location: home; project prefs: default
2/12/2007 3:20:35 PM||General prefs: from SETI@home (last modified 2007-02-11 20:47:06)
2/12/2007 3:20:35 PM||Host location: home
2/12/2007 3:20:35 PM||General prefs: no separate prefs for home; using your defaults
2/12/2007 3:20:35 PM||Reading preferences override file
2/12/2007 3:20:38 PM|SETI@home|Restarting task 20se03aa.22508.29056.859656.3.134_2 using setiathome_enhanced version 515
2/12/2007 3:20:30 PM||Starting BOINC client version 5.8.11 for windows_intelx86
2/12/2007 3:20:30 PM||log flags: task, file_xfer, sched_ops
2/12/2007 3:20:30 PM||Libraries: libcurl/7.16.0 OpenSSL/0.9.8a zlib/1.2.3
2/12/2007 3:20:30 PM||Data directory: C:\\Program Files\\BOINC
2/12/2007 3:20:31 PM|SETI@home|Found app_info.xml; using anonymous platform
2/12/2007 3:20:35 PM||Processor: 1 GenuineIntel Intel(R) Pentium(R) 4 CPU 2.40GHz
[x86 Family 15 Model 4 Stepping 1] [fpu tsc sse sse2 mmx]
2/12/2007 3:20:35 PM||Memory: 1023.48 MB physical, 5.40 GB virtual
2/12/2007 3:20:35 PM||Disk: 128.00 GB total, 118.45 GB free
2/12/2007 3:20:35 PM|SETI@home|URL: http://setiathome.berkeley.edu/; Computer ID: 3083566; location: home; project prefs: default
2/12/2007 3:20:35 PM||General prefs: from SETI@home (last modified 2007-02-11 20:47:06)
2/12/2007 3:20:35 PM||Host location: home
2/12/2007 3:20:35 PM||General prefs: no separate prefs for home; using your defaults
2/12/2007 3:20:35 PM||Reading preferences override file
2/12/2007 3:20:38 PM|SETI@home|Restarting task 20se03aa.22508.29056.859656.3.134_2 using setiathome_enhanced version 515

Wonder why your times are nowhere near mine? Our machines seem quite close. Mine will do 62's in the 7000 second range, even with the generic...Check out my opty times...

@hiamps-

Okay, I've been working with the CPU seconds per credit, comparing my machine to similar ones, such as yours. The CPU seconds per credit seems to go up quickly on the lower credit WUs.

Your Opty is cranking at the rate of about 122 seconds per credit when working with WUs worth about 60 creds. I've got mine down to about 138 seconds per credit for similar sized WUs. This is an improvement over before where my AMD X2 was doing them at the rate of about 155 seconds per credit. You were right, the 1T vs. 2T memory timing parameter seemed to definitely help. Now I've just got to find a way to eliminate another 16 seconds per credit . . .

Brock

---

Batman:

I have been running patched Intel "only" executables with everything from 5.4.x and up to the 5.8.8, which I'm using now, and it has worked perfectly every time.

It has nothing to do with the Boinc client, you are modifying the SETI executable.

If done correctly, the Boinc client, no matter what version, wont know the difference. If you can run the generic client, you can run the patched Intel executable, provided you are aplying it correctly, and using the correct version.

If it doesn't work for you, you are doing something wrong, as there is nothing special about the patched executable, as an executable.

---

I don't think you did anything wrong, Batman.

Personally, I'd suggest that you give your freshly patched science app a benchmark test - I just updated the knabench pack I'm using myself yesterday.

Download that file, uncompress it somewhere, then put your patched app (and any others you want to compare) inside "science_apps/".

Oh yeah, a good idea would be to put an unpatched copy (just rename either the patched or unpatched one) next to the patched one, so you can compare directly.

Next, look in "TestWUs". For quick testing, I suggest moving all but "testWU-4.wu" to the "Reserve" folder. Should be self-explanatory, there's also a text file that describes how to handle things inside "Readme & Licenses". This WU should take less 5-6 minutes for the stock app and 3-4 for optimized ones with any of your Opties.

To run the benchmark, double-click "Knabench-1.43.cmd" in the top-level folder.

Anyway, that package will tell you:
a) how quick your app is compared to stock, or to other optimized apps
b) whether the tested app produces valid results on your host

which is pretty much what you'd want, right?

This way, you can make sure that it's either the app or BOINC, and figure things out from there.

HTH,
Simon.

---

Donate to SETI@Home via PayPal!

Optimized SETI@Home apps + Information

Wonder why your times are nowhere near mine? Our machines seem quite close. Mine will do 62's in the 7000 second range, even with the generic...Check out my opty times...

@hiamps-

Okay, I've been working with the CPU seconds per credit, comparing my machine to similar ones, such as yours. The CPU seconds per credit seems to go up quickly on the lower credit WUs.

Your Opty is cranking at the rate of about 122 seconds per credit when working with WUs worth about 60 creds. I've got mine down to about 138 seconds per credit for similar sized WUs. This is an improvement over before where my AMD X2 was doing them at the rate of about 155 seconds per credit. You were right, the 1T vs. 2T memory timing parameter seemed to definitely help. Now I've just got to find a way to eliminate another 16 seconds per credit . . .

Brock

I had to Manually put my HT up to 800, not sure how much that helped but seems faster than the 400 my Motherboard wanted to use...Other than that I didn't mess with memory much as with my ASUS board I have to clear CMOS with every memory change....

---

Official Abuser of Boinc Buttons...
And no good credit hound!

I've patched the Intel SSE3 Rev 2.0 app from Chicken successfully and let it run on my AMD 64 3500+. I happened to have two identical 0.439194 AR WU's right after another.
SSE 2 generic Chicken app: http://setiathome.berkeley.edu/result.php?resultid=473918377

3 seconds of SSE 2 and the rest with patched SSE 3 (as mentioned above)
http://setiathome.berkeley.edu/result.php?resultid=474339820

A difference of about a minute in favor for the patched SSE 3. Anyone else played with it yet and can share some results? I'll let it run further, problem is that I sometimes have my CPU overclocked, sometimes not. Both results are on the oc'ed CPU though.

---

Well After about a week running the SSE3 it didn't seem to make any difference on my machine, so I just patched the intel SSE2 and have it going and will see if it is any faster. At least I learned something new...The ones that were already in progress now seem to be zipping down in 5 sec intervals...

---

Official Abuser of Boinc Buttons...
And no good credit hound!

Well without much waiting...I would say if you are going to patch an intel version for an Opteron do the SSE2 version. Mine aready is doing faster times.

---

Official Abuser of Boinc Buttons...
And no good credit hound!

hiamps,

it's not that clear of a picture. Depends very much what A64/Opteron model you have - the latest ones like SSE3 better, the earlier X2s like patched SSE2 better, and the single core A64s (like my Winchester) like the generic SSE2 best.

So I can only recommend benchmarking ;)

Regards,
Simon.

---

Donate to SETI@Home via PayPal!

Optimized SETI@Home apps + Information

hiamps,

it's not that clear of a picture. Depends very much what A64/Opteron model you have - the latest ones like SSE3 better, the earlier X2s like patched SSE2 better, and the single core A64s (like my Winchester) like the generic SSE2 best.

So I can only recommend benchmarking ;)

Regards,
Simon.

My old Opty likes the SEE2, My RAC is finally on the rise again. Thanks once more Simon, you make crunching fun! Is there any chance I can help test a new version???

---

Official Abuser of Boinc Buttons...
And no good credit hound!

In a few days, yes ;o)

A new app release is coming up.

Regards,
Simon.

---

Donate to SETI@Home via PayPal!

Optimized SETI@Home apps + Information

In a few days, yes ;o)

A new app release is coming up.

Regards,
Simon.

Like I said...You make crunchin fun...Can't wait!

---

Official Abuser of Boinc Buttons...
And no good credit hound!

Like I said...You make crunchin fun...Can't wait!

OT post:

Hiamps, what kind of critter is pictured in your avatar? P.S I know it's a dog.

tony

deleted

hiamps,

it's not that clear of a picture. Depends very much what A64/Opteron model you have - the latest ones like SSE3 better, the earlier X2s like patched SSE2 better, and the single core A64s (like my Winchester) like the generic SSE2 best.

So I can only recommend benchmarking ;)

Regards,
Simon.

My old Opty likes the SEE2, My RAC is finally on the rise again. Thanks once more Simon, you make crunching fun! Is there any chance I can help test a new version???

@hiamps-

I patched the Intel SSE2 version and tried that- seems to help . . . I'm still trying to figure why your Opty and the other AMD X2s are running about 15% faster than my CPU at the same speed. My L1, L2, and RAM message rates reported in the WUs compare favorably but I'm still lagging about a 1,000 seconds behind regarding the CPU seconds for the same size WUs.

Who's good at solving mysteries?

---

Hi folks,

bet that your patched apps can't match the unpatched newly released Rev-2.2 ones ;o)

A new patchfest may ensue, but hey, it's all in the name of quicker crunching.

Regards,
Simon.

---

Donate to SETI@Home via PayPal!

Optimized SETI@Home apps + Information

Hi folks,

bet that your patched apps can't match the unpatched newly released Rev-2.2 ones ;o)

A new patchfest may ensue, but hey, it's all in the name of quicker crunching.

Regards,
Simon.

We'll find out. I'm already running 2.2 on my Intel P-M and AMD X2. I'll probably upgrade my server tomorrow.

Thanks Simon!

Brock

---