I would like to know how much more productive it is expected to be a HD4870 versus a HD4850.

I would like to know how much more productive it is expected to be a HD4870 versus a HD4850.

20%

As the number of shaders is identical for both cards, the difference comes just from the clock speed, 750 vs 625 MHz.
750/625 = 1.2

The higher memory bandwidth of the HD4870 is of no use for MW. It is not a limiting factor, that's the shader throughput.

I would like to know how much more productive it is expected to be a HD4870 versus a HD4850.

20%

As the number of shaders is identical for both cards, the difference comes just from the clock speed, 750 vs 625 MHz.
750/625 = 1.2

The higher memory bandwidth of the HD4870 is of no use for MW. It is not a limiting factor, that's the shader throughput.

I take it the power of the host computer makes no difference, or does, for example, CPU or RAM in the computer make a difference to the card's performance?

---

I would like to know how much more productive it is expected to be a HD4870 versus a HD4850.

20%

As the number of shaders is identical for both cards, the difference comes just from the clock speed, 750 vs 625 MHz.
750/625 = 1.2

The higher memory bandwidth of the HD4870 is of no use for MW. It is not a limiting factor, that's the shader throughput.

I take it the power of the host computer makes no difference, or does, for example, CPU or RAM in the computer make a difference to the card's performance?

Such a difference should be almost nonexistent as long as at least 2 WUs run concurrently. That enables the possibility to hide the CPU part of one WU during the GPU computations of another one.

the gpu app has been running a little funny today Normally it only runs one at a time this morning it was running 4-5 at a time and bogging down the system quite a bit. I rebooted and that seemed to fix it, for a little while but now it's not running at all, and has a message on a wu saying waiting to run (.50 cpus). Is it waiting for my other projects to finish up on the cpu for some reason even though it can run while they are still going?

the gpu app has been running a little funny today Normally it only runs one at a time this morning it was running 4-5 at a time and bogging down the system quite a bit. I rebooted and that seemed to fix it, for a little while but now it's not running at all, and has a message on a wu saying waiting to run (.50 cpus). Is it waiting for my other projects to finish up on the cpu for some reason even though it can run while they are still going?

These scheduling problems are caused by the fact that the BOINC Client does not know the app is running on the GPU. It is scheduled like a normal app running on the CPU.

i left it at 9.1 but took your advice of upping the fan speed. that's brought it down from ~83c to ~77c.

and it made it through the night without locking up. So it could be that it was overheating. But I had also upgraded to version .19. So, was there a change in there that corrected the problem?

either way, v.19 is working fine on an i7 and hd4870 with 512m

Besides the CPU detection nothing changed between 0.17 and 0.19. So maybe really a temperature problem.

But I have seen your crunch times are slightly on the high side. This could be caused by running too many WUs concurrently on the GPU. At a certain point the RAM on the graphics card is not sufficient for the number of WUs taking space there. Before it errors out (when even more WUs would be crunched), it slows down (probably some swapping over PCI-Express happens). And with 16 WUs it is getting already a bit crowded on a 512MB card.
Another reason for the higher times could be that the card runs downclocked in a power saving mode. Maybe you should check the clock speed of the card.

Furthermore you may think about attaching to a second BOINC project with that i7. This will reduce the number of MW WUs that are running a he same time, but not the throughput. You will still finish the same number of WUs per hour even with less concurrently running WUs. In fact, it could even rise in your case. Furthermore your CPU cores wouldn't be idling that much ;)

I enabled CPDN and ABC but MW always ran 16 apps, then would switch to 8 ABC, never sharing - it was like how the Cylons and Humans can't get along, it always had to be one or the other never both.

But with 16 MW's running it would eventually lock up. It really does look like a ram problem, the 512 really can't handle a constant run of 16 Apps of this length. Though I didn't have trouble earlier with the shorter apps.

So, what I finally did was reset boinc to only run with 6 CPU's max. Now with only 12 wu's running at a time, the GPU seems much happier and doesn't lock up. And the actual crunch times are significantly shorter with only the 12. 2 - 3 mintues compared to the 8 or 9 when running 16. And as you pointed out, I end up processing more wu's.

- it was like how the Cylons and Humans can't get along, it always had to be one or the other never both.

I've been running the HD4850 on version 0.19 for just over 24 hours now and the phrase 'running like the clappers' comes to mind ;) I've tried to share with CPDN, but it instantly drops from a rapid 4 x 0.5 CPU to a very slow 2 x 0.5 CPU. To be honest I can run CPDN on another box and the 4 x 0.5 clappers is more than enough throughput for me on this one box. However, if a way to share with the Cylons can be found, it would be nice ...

---

All right, got the GPU app working, under Windows 7 no less. Two problems: BOINC was using a protected install, and the DLLs for stream were missing. First was easy to correct with a reinstall, and the DLLs I tried grabbing from Vista x64, worked like a charm. So in exchange for some graphics performance, I get 4 ABC and 4 MW WUs running all at the same time. Great app Cluster Physik. ;)

---

Please, I need to know if the Win32/ATI/0.19 application will work under Vista 32 bits O.S. I have a just-bought hd4850 and I need to know it before I open the box breaking the seals

I have BOINC 6.4.5 installed as a protected service on XP64 (domain member), and it works a charm with Catalyst 8.12 (even 9.1) and the GPU app 0.19 (all hail Cluster Physik!).

However, I just tried to start BOINC Manager locally (rare event, I normally rely on BoincView from another computer rather than that useless PoS), and immediately this happened:

<message>
Incorrect function. (0x1) - exit code 1 (0x1)
</message>
<stderr_txt>
Running Milkyway@home ATI GPU application version 0.19 by Gipsel
CPU: Intel(R) Xeon(R) CPU           X3360  @ 2.83GHz (4 cores/threads) 4.00399 GHz (332ms)

No compatible GPU found!

</stderr_txt>

After i killed Boing Manager and rebooted, things went back to normal again. Repeating the excercise led to an identical outcome. No biggie, bit it did get my attention.

---

All right, got the GPU app working, under Windows 7 no less. Two problems: BOINC was using a protected install, and the DLLs for stream were missing. First was easy to correct with a reinstall, and the DLLs I tried grabbing from Vista x64, worked like a charm. So in exchange for some graphics performance, I get 4 ABC and 4 MW WUs running all at the same time. Great app Cluster Physik. ;)

Having discovered a stopwatch on my mobile phone I ran some tests and find that I can indeed run another project together with MW. I quite agree - Great app Cluster Physik. ;)

---

I previously tried Catalyst 9.1 for my HD4850 which crashed and left the MW tasks going snail-pace.

I also tried Catalyst 9.2 with the renames in Windows/System32, but it again crashed and snail-pace.

Reinstalled Catalyst 8.12 and everything is rippling again nicely ;)

---

It worked great on Vista64, but on Xp32 I get always the same message : not requesting tasks ... the only solution I found was to suspend or delete the other projet (World Comm Grid).

I think this is a priority problem (100 for each, theorically 50% each).

Edit : I tried 100/900 (WCG/MW), but now it's running only WCG because of priority time... I try to suspend WCG.

Edit 2 : With only MW project, I get some workunits and then :

24/02/2009 08:53:11 Milkyway@home Sending scheduler request: Requested by user.
24/02/2009 08:53:11 Milkyway@home Reporting 3 completed tasks, requesting new tasks
24/02/2009 08:53:16 Milkyway@home Scheduler request completed: got 0 new tasks

---

Star Wars BOINC Team

Strange error message, also had 8 MW tasks with this error:

<message>
There are no child processes to wait for. (0x80) - exit code 128 (0x80)
</message>

Not sure if it is significant, or something to do with my computer going nuts on running the hard drive, or this is what caused my computer to run the hard drive ...

A reboot and some hand turning of tasks running and I seem to be back on the air ...

Not sure what it means, if it means anything, luck of the draw, bad day at black rock ...

Not sure what it means, if it means anything, luck of the draw, bad day at black rock ...

Bad day at black rock. If you're running the ATI then these tasks are over in a blink of an eye. Let them go. We cherish each one, I know. But there are many others to greet and to race by - in the blink of an eye...

;)

---

Not sure what it means, if it means anything, luck of the draw, bad day at black rock ...

Bad day at black rock. If you're running the ATI then these tasks are over in a blink of an eye. Let them go. We cherish each one, I know. But there are many others to greet and to race by - in the blink of an eye...

;)

Hmm, well, I don't know about that ... I blink slowly, but not THAT slowly ... :)

But you are right, I am too anal about losing the science ...

Not sure what it means, if it means anything, luck of the draw, bad day at black rock ...

Bad day at black rock. If you're running the ATI then these tasks are over in a blink of an eye. Let them go. We cherish each one, I know. But there are many others to greet and to race by - in the blink of an eye...

;)

Hmm, well, I don't know about that ... I blink slowly, but not THAT slowly ... :)

But you are right, I am too anal about losing the science ...

I know what you mean. I find it hard to micro manage so many machines, and especially have that 'Doh!' feeling when I find I've crunched the same CPDN WU and it ain't gonna give me no more credit the second time. I have to ditch some part run WUs 'just in case' I already finished them off somewhere else.

Oh the joy of short MW tasks :)

---

Well, I don't have that many systems any more ... the wife has been on me to cut back and I have been doing so ...

I am only running 6 systems now and actually should only be running 5 (I have had a linux system up for a month or so now from my graveyard) ... but, it has VP almost to 20K and WEP over 10K so it is about time to shut it down for the summer (The two projects I am running there)...

But could you please test that it runs at all?

Here's an update on the double-GPU HD4870X2 vs the single-GPU HD4870/1GB.

HD4870/1GB runs at its highest configured clock speeds when under load from the MW app. Partly OC'ed to 800/900 MHz (gpu/mem) this means 5-8 seconds per WU.

HD4870X2 runds at its idle clocks (500 MHz gpu) even with one GPU under load from MW, which means 8-12 seconds per WU. It seems that giving half the card something to do isn't enough to "wake it up" from idle/2D clocks. But yes, the GPU apps do work on this card as well, albeit somewhat slower than on its lesser sibling.

---