I have been running a GTX 460 doing two units at a time. I decided to put my old (non-Fermi) 9800 GTX+ board in but did not remember to go back to single unit runs.

Now here's the deal, both cards show progress in 2 units. Honest to God the 9800 GTX+ is doing two units at a time and each unit is showing different progress as you would expect.

They are running at a little less than half the rate of the two that the 460 is cranking them out . . . say 45-50 minutes t completion.

Put that in your pipe and smoke it.

Any ideas???

---

Man's future lies in the stars, not on Earth. It is each successive generation's responsibility to humanity to expand the knowledge and understanding of our Universe so that we may one day venture forth to meet our neighbors.

Houston, Texas

If you have one of the 1gig 9800 cards, this does not sound surprising.
It may also work on a 768meg.. But not on a 512Meg.

My 9600 is still plugging along.

---

Janice

I thought anything less than a fermi would basically not be able to run the WU's efficiently. making the run times actually longer than the single run WU's

---

In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope

The 9800 is a great old card and, as you've proved, can run two at a time but, as Skildude pointed out it could be a lot slower than what just one at a time will do. Just make sure you don't ask too much of it and keep your monitor on the 460. Also watch the temps coming out of it, make sure it doesn't burn itself out trying to please you.

---

PROUD MEMBER OF Team Starfire World BOINC

That's right. I'm always careful with my oldest GPU. They perform great if properly installed and taken care of.

---

rOZZ
Music
Pictures

Well, it looks like the old 9800 GTX+ will run two units in about 1 hour to an hour and ten minutes and my GTX 460/768 is running 2 units in 20 minutes as it has been doing.

GPU temps are 65 and 67 and usage hovers around 90%.

I should add that I am running an AMD 955 BE on one of those MSI Fuzion boads that uses the Lucid HydraLogic circuitry and there appears to be some load sharing going on between the two boards on occassion . . . don't know it that is real or not.

Anyway, things look stable for now. Definitely can feel the additional heat from the second board.

---

Man's future lies in the stars, not on Earth. It is each successive generation's responsibility to humanity to expand the knowledge and understanding of our Universe so that we may one day venture forth to meet our neighbors.

Houston, Texas

do you have any work completed on the 9800 without the doubling to compare. are they finishing at the same times as before?

---

In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope

9800GTX+ should be able to finish an average unit in 15-17 minutes.

---

looking at his pending I see work getting done in 22-25 minutes. Not sure if this is the 9800 or 460 but it appears that you are actually getting more out of the GPU. I have to assume that the BOINC process looks at the GPUs as being 2 460's so we can't tell which is running the work

---

In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope

looking at his pending I see work getting done in 22-25 minutes. Not sure if this is the 9800 or 460 but it appears that you are actually getting more out of the GPU. I have to assume that the BOINC process looks at the GPUs as being 2 460's so we can't tell which is running the work

If you look at the stderr.txt output it tells you:

<core_client_version>6.10.58</core_client_version>
<![CDATA[
<stderr_txt>
setiathome_CUDA: Found 1 CUDA device(s):
Device 1: GeForce GTX 460, 738 MiB, regsPerBlock 32768
computeCap 2.1, multiProcs 7
clockRate = 810000
setiathome_CUDA: CUDA Device 1 specified, checking...
Device 1: GeForce GTX 460 is okay
SETI@home using CUDA accelerated device GeForce GTX 460
Priority of process raised successfully
Priority of worker thread raised successfully

Claggy

I looked at 30+ Wu's and all are labeled 460. I thought BOINC had a problem with multiple GPU's and it would label all the GPU's as the first that it sees. Or has this changed?

---

In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope

I looked at 30+ Wu's and all are labeled 460. I thought BOINC had a problem with multiple GPU's and it would label all the GPU's as the first that it sees. Or has this changed?

Boinc might have a problem with it, but the app hasn't, Cuda apps will report what Cuda GPU's are there,

See the stderr.txt outputs from these top hosts:

Stock Cuda app

x32f Cuda app

V12 Cuda app

and here's a host with a mixture of GPU's: resultid=1778817269

Claggy

Guys, he just started this thread this morning. He probably hasn't reported any of the new work yet. Give it a while and lets see if it changes.

---

PROUD MEMBER OF Team Starfire World BOINC

So far it's looking like an over all boost of 33% in my production.

With the 460 I was averaging 2 units every 20 minutes, or 6 units/hr.

The 9800+ is averaging 2 units in about 56 minutes, say 2/hr, so I've gone from 6/hour to 8/hour.

I can live with that. :) Beats just having the card sit around gathering dust.

I had tried just running one unit per card a while ago, but the 460 by itself running 2 at a time was faster than 1/card using both cards.

Now the only question is if it is generating good results and for that I have to wait until they can be reported in.

Boy, this is exciting . . . okay, so I'm easily entertained. :)

GPU 0 (the 460) is running 63C at 90%
GPU 1 (the 9800) is running 65C at 84%

Keeping both fans on manual at 80%.

Hey, I even watched a movie while they were both running with no difficulty.

---

Man's future lies in the stars, not on Earth. It is each successive generation's responsibility to humanity to expand the knowledge and understanding of our Universe so that we may one day venture forth to meet our neighbors.

Houston, Texas

What clocks your 460 is running with?
I see very often for 460s clocks like 810000. which should be shader clock for low power 3D mode, when core is running on 405 MHz.
My 460s show their real shader clocks, 1620 MHz(1620000).
Yeah, with Fermi cards its possible to use desktop with no stuttering while doing SETI.

---

The 460/768 is running at 850/1700 core/shader and 2000 memory . . . pretty much maxed out on OC'ing

The 9800GTX+ is running stock at 738/1836 and 1100.

Every thang is rock steady, y'all.

---

Man's future lies in the stars, not on Earth. It is each successive generation's responsibility to humanity to expand the knowledge and understanding of our Universe so that we may one day venture forth to meet our neighbors.

Houston, Texas

Someone please explain. Since the last outage I've lost around 800 RACs. Will this be recovered in due course when the servers are fully operational. In other words is this a normal occurrence. Member sin December 8, 2011.

---

Someone please explain. Since the last outage I've lost around 800 RACs. Will this be recovered in due course when the servers are fully operational. In other words is this a normal occurrence. Member sin December 8, 2011.

Pretty much, I lost about 3200 but my pending is up quite a bit as well.

This is after getting all my work uploaded and reported.

---

Same here I lost about 7k last check before outage was a touch over 31k. However that being said now that everything is uploaded I have about 2-3x more pendings than I had, so I imagine it will rebound for possibly another record day!

---

Traveling through space at ~67,000mph!

It's happened again. Why should I go on "crunching" tasks when every time there's an outage my RAC's slip backwards. Very frustrating. I've got upto 5 days tasks to run should I say no more. Power usage (carbon credits) are starting to hurt me.