To Jord and Fluffy-
Everything is ok now. I called a friend at local uni computer dept and he came round - recommended and carried out a complete removal of BOINC and Rosetta and reinstallation.
He thinks there is something in BOINC 5.8.8 caused the problem but did not check [after I described the symptons], but says my web browser is perfect, same as installation 3 months ago!! IE7 is also good.
At any rate am pleased to be up and running. Sorry about the lost WUs.
Thanks for all your help.

---

further to my prev message[36150] - problem was solved by reverting to previous BOINC version 5.4.11. Seems as if 5.8.8 was the problem.

---

On Core 2 , i get: GenuineIntel
Intel(R) Core(TM)2 Quad CPU @ 2.66GHz [x86 Family 6 Model 15 Stepping 7] [fpu tsc pae nx sse sse2 sse3 mmx]

I made it bold in your reply.

He said SSSE3 ;)

---

He said SSSE3 ;)

Ah true. I misread that. 't was early in the morning here as I read that. ;-)
But it's just a revision of SSE3. Linky.

---

CPU detection in new BOINC is missing SSSE3
On Core 2 , i get: GenuineIntel
Intel(R) Core(TM)2 Quad CPU @ 2.66GHz [x86 Family 6 Model 15 Stepping 7] [fpu tsc pae nx sse sse2 sse3 mmx]

I know for sure (hehehehe) that my Processors include SSSE3.

Somebody can flag it to the boinc people?


who?

Who? BOINC itself is not at fault, it is Windows VISTA. It does not detect SSSE3 in the kernel or maybe it has no call to ask for it. (3xS's)

Strangly, in Windows XP and although Windows XP has the command to ask for SSE3 it does not seem to detect it. (that's 2xSSE3, there is no call to ask for SSSE3)

All BOINC does is ask the OS what is supported,
For Windows this is using the
IsProcessorFeaturePresent()

Linux I think it is by reading /proc/cpuinfo
and for MacOS-X last I heard they where going to use a sysctl() call. It'll be in the code somewhere.

---

Team mauisun.org

Looks like the new release also honors the General Preference for limiting use of your network bandwidth. This will be great for those on dialup that want to do other things with their internet connection while BOINC is doing a routine update.

---

Add this signature to your EMail:
Running Microsoft's "System Idle Process" will never help cure cancer, AIDS nor Alzheimer's. But running Rosetta@home just might!
https://boinc.bakerlab.org/rosetta/

I don't even want to repeat what I had to go through to get this new version of Boinc to work on computer. Suffice to say, to anyone who is thinking about installing the new Boinc, use the "No New Work" function on your current client so the client doesn't get any new work. Let what is running run till it is finished, let it report, then close the Boic client. Then install the new one. Then restart your computer, and make sure you look very closely at the selections under your "Activity" button at the top, to make sure it is not set on "Suspend" and "Network Activity Suspend."
I lost a work unit that was 4 hour in because the new client didn't want to restart it. Or connect for new work after I aborted it. I had to restart my computer twice because it wouldn't fetch new work. What a pain. And the basic view of the client looks like crap.

And now Who's MIPs are off. How is he going to win the Golden Chromosome Award?

One would have to want to update pretty badly to do all of that and still lose.

---

dag
--Finding aliens is cool, but understanding the structure of proteins is useful.

The OS does not have to detect an Instruction set, it is simpler than this:

to detect an instruction set as SSEx, you need to use a TRY/EXCEPT methode,

something like

try
{
lddqu xmm0,[eax] //(SSE3 instruction_
}
except
{
SSE3=1;
}


Got the idea?
Ofcourse, you have to set EAX to the right value.
By doing so, you do it in a way that does not requires AMD or Intel custom code.
Best way to detect an instruction set, is to try it :)

who?

I think that is the way they may be going, I've not followed the 5.9.x series code for a while, but they specifically (well Rom Walton) didn't want to go looking for instuctions sets since the code would constantly need updating and a new release ever time there was a new one to add. It's there more for information and only one project at the moment probably needs it (docking@home) since they just noticed not all Celerons are the same ;-). Actually thinkign about it, the stepping/model etc. Was added because of that (again just whatever the OS reported), the instruction set just got moved into that string from it's own string in 5.6.x series.

I think in their opinion if a project needed the instuction sets they should check for it themselves. Though if you have some information and better ways they could incorporate it just email Rom Walton or post to the boinc_dev lost http://boinc.berkeley.edu/email_lists.php

The OS does not have to detect an Instruction set, it is simpler than this:

to detect an instruction set as SSEx, you need to use a TRY/EXCEPT methode,

something like

try
{
lddqu xmm0,[eax] //(SSE3 instruction_
}
except
{
SSE3=1;
}


Got the idea?
Ofcourse, you have to set EAX to the right value.
By doing so, you do it in a way that does not requires AMD or Intel custom code.
Best way to detect an instruction set, is to try it :)

who?

---

Team mauisun.org

It seems that with tonight's tweaking the default Boinc credit system is back in place. Granted credit=claimed credit, and is based on reported FLOPS. A short-term mistake while the team settles things down, I hope.

---

thanks for catching that. it's fixed now.

Some of my Rosetta forum settings were reset. I used to have it where it showed newer messages first, now it is showing older messages first. But this could be something I did on my end.

---

Paydirt, the forum preferences weren't all preserved :( The issue is discussed here.

---

Rosetta Moderator: Mod.Sense

In 5.8.11 and 5.8.14
Don't downlaod NEW work's after compliting.
only after compliting all work's in Cache!

---