Hi, I've downloaded BOINC and joined CPDN yesterday but the graphics for CPDN workunit don't appear. SETI@Home works fine but when I click "Show graphics" when CPDN is working and it's selected on Work tab, the BOINC manager window just flashes and nothing happens (no window...). When I run the BOINC screensaver, SETI@Home displays everything but CPDN does nothing again (the screen is all black). Simulation seems to run (it's at 0.30% by now).

I've tried to search these forums but found nothing useful (ussually "check drivers" etc, when there's something similar, it ends with "I've got it working, thanks all" with no details).

My system is M$ Windoze XP (I'd love to run it under Linux but I can't get my modem up, there're no drivers for it yet...) with ATI Radeon 8500 and latest Catalyst drivers. Am I missing something?

---

>Am I missing something?

Probably not. The visualisation in CPDN is a far more sophisticated affair than other DCs, and I'm only sorry that you can't see it. It's given some people trouble before, and the man behind it, Tolu, put a lot of work into an upgraded BOINC version, but for the last few months he has been standing in as chief software engineer for the project trying to cope with the change in the BOINC client, routine problems including the servers, and develop a new version of the experiment. He'd dearly love to sort out the visualisation but I'm afraid that if he gave his time to that the rest of us would probably have something to say about it.

All is not lost, though. There is an advanced version of the visualisation being developed out of house which with luck should be out for all platforms before much longer. You may have to wait for that.

See also <a href="http://climateapps2.oucs.ox.ac.uk/cpdnboinc/forum_thread.php?id=2163">this thread</a>.

---

It seems that it may be a BOINC, rather than CPDN problem - see <a href="http://setiweb.ssl.berkeley.edu/forum_thread.php?id=12948">this thread on the S@H board</a>.

---

&gt; It seems that it may be a BOINC, rather than CPDN problem -

From what I've seen in most of these threads there is a problem with some of the ATI drivers. Try downgrading your video drivers.

---

BOINC WIKI

BOINCing since 2002/12/8

&gt; He'd dearly love to sort out the visualisation but
&gt; I'm afraid that if he gave his time to that the rest of us would probably have
&gt; something to say about it.

Well, spending an hour to add some init checks with log messages and releasing it as debugging version wouldn't hurt. Hunting the problem from DX diagnostics like S@H admins do is the worst approach I can imagine. Log messages can tell me what's really wrong.

---

next_ghost
Modifying the hadsm code is not a trivial matter. The source is over a million lines of fortran code for a super computer.
And BOINC is a Berkeley Uni program, not Oxford Uni.
And lots of people are NOT having problems. Perhaps because we don't have an ATI card.

Les

It may be worth further complicating the discussion by reiterating that by no means all people with ATI cards are having problems. I'm using BOINC 4.26 (withdrawn but it's OK for me)with an ATI Radeon 8500 and the Microsoft certified driver (6.14.10.6462). I don't use the graphics a lot but they haven't crashed.

---

Thats what I like, Andrew. More complications. :)

Perhaps we should ask users WHICH ATI card they use, and what driver. It may show a better pattern than just ATI.

I'm just glad that I went for a NVIDIA card, otherwise Murphy would have gotten me. Don't use the viz much either.

Les

&gt; next_ghost
&gt; Modifying the hadsm code is not a trivial matter. The source is over a million
&gt; lines of fortran code for a super computer.
&gt; And BOINC is a Berkeley Uni program, not Oxford Uni.
&gt; And lots of people are NOT having problems. Perhaps because we don't have an
&gt; ATI card.
&gt;
&gt; Les

I'm not asking to modify the entire hadsm code. What I'm asking for is to add a few check lines of code to already working init checks that terminate graphics before something is actually drawn. OpenGL/D3D init takes at worst 200 lines of code WITH those checks. It'll do everything it does now but it will at least say WHERE it actually did exit. If u still don't understand, I want this:

if(blahblah() == NULL)
{
exit(1);
}

turned into this:

if(blahblah() == NULL)
{
log("Critical error: blahblah() returned NULL, terminating graphics...");
exit(1);
}

---