Since it looks like 5.2.2 has been released, would it be possible to get the optimizing team (Ned, TMR, Trux) to come out with some CPU specific boinc.exe's.. Thanks guys for your help!

---

- cJ

Since it looks like 5.2.2 has been released, would it be possible to get the optimizing team (Ned, TMR, Trux) to come out with some CPU specific boinc.exe's.. Thanks guys for your help!

Why not 5.2.4? It looks stable judging from the cvs "stable" tag. I pulled the source just after reading your post. :)

cvs -z3 -d :pserver:anonymous@alien.ssl.berkeley.edu:/home/cvs/cvsroot checkout -r stable boinc

boinc-524-mmx.zip

boinc-524-sse.zip

boinc-524-sse2.zip

(no sse3 version because it's almost the same as sse2.)

PS: I haven't tried these yet, because I downloaded & built them just now. These will work with non-intel processors, but benchmark (whetstone) may be lower because of Intel's preference in its math library. I cannot hack because it will violate Intel's EULA, and that I have only Intel processors.

---

Luckiest in the world. WMD = Weapon of Mass Distraction.
Click this table.

The SSE2 version is corrupted.

Since it looks like 5.2.2 has been released, would it be possible to get the optimizing team (Ned, TMR, Trux) to come out with some CPU specific boinc.exe's.. Thanks guys for your help!

Why not 5.2.4? It looks stable judging from the cvs "stable" tag. I pulled the source just after reading your post. :)

cvs -z3 -d :pserver:anonymous@alien.ssl.berkeley.edu:/home/cvs/cvsroot checkout -r stable boinc

boinc-524-mmx.zip

boinc-524-sse.zip

boinc-524-sse2.zip

(no sse3 version because it's almost the same as sse2.)

PS: I haven't tried these yet, because I downloaded & built them just now. These will work with non-intel processors, but benchmark (whetstone) may be lower because of Intel's preference in its math library. I cannot hack because it will violate Intel's EULA, and that I have only Intel processors.

---

The SSE2 version is corrupted.

sorry, now it's fixed.

These will work with non-intel processors, but benchmark (whetstone) may be lower because of Intel's preference in its math library.

Can someone put benchmark figures up for before and after??

But you know what?? The boinc client just handles the communications for the project. The project client program does the actual crunching so having an optimized Boinc will have little if any effect on crunch times.

---

Boinc....Boinc....Boinc....Boinc....

The SSE2 version is corrupted.

sorry, now it's fixed.

I downloaded the SSE2 version again but when I place it over a clean 5.2.2 install I get a failed to initialize error when BOINC tries to start. The SSE version runs ok.

---

I closed 5.x and then installed 5.2.2. The only options that came up are REPAIR or UNINSTALL. I hit REPAIR and it installed fine but didn't recognize a version change and didn't run benchmarks automatically.

---

-----

The SSE2 version is corrupted.

sorry, now it's fixed.

I downloaded the SSE2 version again but when I place it over a clean 5.2.2 install I get a failed to initialize error when BOINC tries to start. The SSE version runs ok.

it works perfectly on my side. I put it in an empty directory and launch boinc.exe. I could attach to alpha project and it began crunching.

On another box, I overwrote over 4.72 and it works fine.

---

Luckiest in the world. WMD = Weapon of Mass Distraction.
Click this table.

it works perfectly on my side. I put it in an empty directory and launch boinc.exe. I could attach to alpha project and it began crunching.
On another box, I overwrote over 4.72 and it works fine.

Unzipping SSE2 into a blank directory and running the new boinc.exe produces the error: "The application failed to initialize properly (0xc0000005). Click on OK to terminate the application." This message also appears when the SSE2 version is installed over a clean v4.72 or v5.2.2. The SSE version you've provided works just fine in all cases. This is on an SSE2 enabled Pentium M (Dothan).

---

it works perfectly on my side. I put it in an empty directory and launch boinc.exe. I could attach to alpha project and it began crunching.
On another box, I overwrote over 4.72 and it works fine.

Unzipping SSE2 into a blank directory and running the new boinc.exe produces the error: "The application failed to initialize properly (0xc0000005). Click on OK to terminate the application."
This is on an SSE2 enabled Pentium M (Dothan).

Then check .dll file dependency using dependency walker or the likes. Otherwise I have no idea, since sse2 version is working perfectly on my boxes either in a clean directory or over 4.72. All the necessary .dll files are provided in older (maybe 4.45 or something) boinc core.

---

Luckiest in the world. WMD = Weapon of Mass Distraction.
Click this table.

Same error here, Tetsuji...
The sse2-version wil not run (giving the exact same error). The sse-version is working fine.
Running it on an intel p4, 2.4 Ghz (ht, sse2)

Regards,

fic.

---

If we upgrade from 4.4.5 to the 5.2.2 boinc, can we use the existing optimized seti clients, such as YAOSCW-P-r8.1?

Or will those optimized clients only work with 4.4.5? Do we need all new clients? (And all new benchmarking, to determine which to run?)

Thanks!
Don

---

If we upgrade from 4.4.5 to the 5.2.2 boinc, can we use the existing optimized seti clients, such as YAOSCW-P-r8.1?

Or will those optimized clients only work with 4.4.5? Do we need all new clients? (And all new benchmarking, to determine which to run?)

Thanks!
Don

You can keep on using the same optimized client...

Greetings,
Sander

---

it works perfectly on my side. I put it in an empty directory and launch boinc.exe. I could attach to alpha project and it began crunching.
On another box, I overwrote over 4.72 and it works fine.

Unzipping SSE2 into a blank directory and running the new boinc.exe produces the error: "The application failed to initialize properly (0xc0000005). Click on OK to terminate the application." This message also appears when the SSE2 version is installed over a clean v4.72 or v5.2.2. The SSE version you've provided works just fine in all cases. This is on an SSE2 enabled Pentium M (Dothan).

I get the same error message. The SSE version works nicely. I never had these problems with the 4.45 SSE2 version.
Bye Dee

it works perfectly on my side. I put it in an empty directory and launch boinc.exe. ...

Unzipping SSE2 into a blank directory and running the new boinc.exe produces the error: "The application failed to initialize properly (0xc0000005). Click on OK to terminate the application."
...

I get the same error message. The SSE version works nicely. I never had these problems with the 4.45 SSE2 version.
Bye Dee

Got the same error message in a blank dir after stopping boinc 4.72.
Didn't want to risk my work just for a test, so didn't overwrote my 4.72

Andy

Edit:
With the SSE version noc problem in the dir.
got following benchmarks (SSE):
float: 2210.210
int: 5562.994
sum: 7773,204

with the 4.72 SSE2
float: 3012.386
int: 4962.645
sum: 7975,031

Waiting for a 5.x SSE2 version to test again.

Andy

---

Want to know your pending credit?


The biggest bug is sitting 10 inch in front of the screen.

Thanks Sander.

I can't find any info on upgrading vs. installing.

Does it work to just install over the same directory as 4.4.5? Will it keep all the WUs?

Will it keep using the optimized client?

(Yes, I'll make a backup copy of the dirs, just in case...)

Thanks,
Don

---

I just upgraded to 5.22 and tried to put in the SSE2 version and got the same errors as above. went back to 4.72 for now until this one is fixed.

---

I can't find any info on upgrading vs. installing.

Does it work to just install over the same directory as 4.4.5? Will it keep all the WUs?

Will it keep using the optimized client?

(Yes, I'll make a backup copy of the dirs, just in case...)

Each version can be different, but it seems you can install over the existing directory just fine. If you like to play it safe, you can completely uninstall BOINC and then install the new 5.2.2 version. (Major version upgrades sometimes require this and it may prevent orphaned files or unforseen version interactions from taking place.) Uninstalling will leave your BOINC folder intact (don't delete it!) and all of your project information will be kept, including WU's in progress and in queue. It will continue to use any optimized project applications that you have in place. Install the new version, fire it up and it should pick up right where the old version left off after benchmarks have been run.

Backing up is always a sensible thing to do just in case something goes wrong!

---

I upgraded a few hours ago.

Some users have indicated that it is possible to install the new version over the old version. I prefer to upgrade a bit differently. This is what I did for a successful install.

I first stopped Boinc.

Made a backup of the Boinc directory.

Uninstalled Boinc version 4.45.

Installed Boinc version 5.2.2.

Restarted Boinc.

Everything is running fine at this point. Have sucessfully downloaded new work for Seti and Protein Predictor. And Climate prediction keeps on chugging along...




Your mileage may vary.

---

Since it looks like 5.2.2 has been released, would it be possible to get the optimizing team (Ned, TMR, Trux) to come out with some CPU specific boinc.exe's.. Thanks guys for your help!

Why not 5.2.4? It looks stable judging from the cvs "stable" tag. I pulled the source just after reading your post. :)

cvs -z3 -d :pserver:anonymous@alien.ssl.berkeley.edu:/home/cvs/cvsroot checkout -r stable boinc

boinc-524-mmx.zip

boinc-524-sse.zip

boinc-524-sse2.zip

(no sse3 version because it's almost the same as sse2.)

PS: I haven't tried these yet, because I downloaded & built them just now. These will work with non-intel processors, but benchmark (whetstone) may be lower because of Intel's preference in its math library. I cannot hack because it will violate Intel's EULA, and that I have only Intel processors.

I don't see 5.2.4 anywhere other than Mac and Linux.. Where is the install for 5.2.4 windows? I don't see it on ?dev=1 or on /dl

---

- cJ