Is it worthy to crunch on my amd 1090 3.6 ghz? How much credit will it make?
Also, can you suggest a backup projects for my machine: AMD 1090 3.6, ATI 6970.

Thanks

Up

Thought someone else might chime in, but since no one has...
I'd say do MW on your 6970 and attach to another project for the CPU. MW will run great on the 6970, but there are other projects better suited to any CPU in my opinion.

HTH,
-Dave

---

If you are looking for a CPU application, go for AQUA. They recently resolved many issues with their current main application (ROQS), including a credit revamp.

At the new rates an AMD Phenom 9XX would expect to do an ROQS in around 10-12 hours for just under 31,000 credits (note that the application uses all available cores, dont calculate that as 31,000 per core). Works out in the region of 60,000 a day - a 1090 probably close to 70,000.

In the past with AQUA, you would loose more credits overall running it as well as another Project GPU Application, than you would gain running GPU only elsewhere. However with the new credit scale from a few days ago, you would only loose around 15000-18000 a day off a GPU app when running an AQUA CPU application, but gain 70,000 (ish) from the AQUA CPU app for a 1090T. So now you have a net gain, and is worth doing.

So if credits are the main driver, AQUA CPU is now a BOINC no brainer, and the best around in a CPU credit sense, as its credit scale more than makes up for the hit on a concurrent GPU Application when running it.

Horses for Courses as they say :)

Regards
Zy

Thanks a lot

Will give it a shot

FWIW, i don't recommend using your 1090 on AQUA, since it will interfere with MW and slow down your GPU, which is a fact.

Besides that, the RAC posted by some that your 1090 could achieve over at AQUA is far from reality. Instead I'd suggest using it over at primegrid using the PPS Sieve app where it will get the best RAC possible.

---

Join Support science! Joinc Team BOINC United now!

FWIW, i don't recommend using your 1090 on AQUA, since it will interfere with MW and slow down your GPU, which is a fact.

Besides that, the RAC posted by some that your 1090 could achieve over at AQUA is far from reality. Instead I'd suggest using it over at primegrid using the PPS Sieve app where it will get the best RAC possible.

I extreamly second that, Crunch3r. Aqua Roqs will need at least 2 free cores to feed the gpu( as in my case on both my i7 and AMD) taking a big hit on ANY gpu'ing. PPS sieve, on my AMD P-IIX4 9650 avg. 24/30 hours, reguardless you get 4500. Collatz is a good go-to for cpu. Im thinking of building a cheap i7 960 just for Aqua/Roqs? I like the Quantum computer aspect. By the way, what can I expect frm 2 5870? Stock clock, two differnt rigs. Im using TThrotle at 130 deg. F, runs avg. 75% day 100% at nite. They still finish around 2 min.

---

FWIW, i don't recommend using your 1090 on AQUA, since it will interfere with MW and slow down your GPU, which is a fact.

Besides that, the RAC posted by some that your 1090 could achieve over at AQUA is far from reality. Instead I'd suggest using it over at primegrid using the PPS Sieve app where it will get the best RAC possible.

I extreamly second that, Crunch3r. Aqua Roqs will need at least 2 free cores to feed the gpu( as in my case on both my i7 and AMD) taking a big hit on ANY gpu'ing. PPS sieve, on my AMD P-IIX4 9650 avg. 24/30 hours, reguardless you get 4500.

Yes, that's what i was talking about. AQUA's ROQS is a dog and has some serious issues. It doesn't utilize all cores while running. On a V8 xeon, it uses less than 4 cores most of the time in the hot loop... (between 49 to 55 % progress)

Collatz is a good go-to for cpu. Im thinking of building a cheap i7 960 just for Aqua/Roqs? I like the Quantum computer aspect. By the way, what can I expect frm 2 5870? Stock clock, two differnt rigs. Im using TThrotle at 130 deg. F, runs avg. 75% day 100% at nite. They still finish around 2 min.

Collatz on CPU using a 64 bit OS is almost as close to PPS Sieve credit wise and works even better along with GPU apps like MW, having no effect on MW GPU times at all.

---

Join Support science! Joinc Team BOINC United now!

That used to be true with AQUA, the hit on the running gpu app negated the gain on running the AQUA cpu app, but thats no longer the case.

They doubled the credit allocation for ROQS, resulting in 50,000 - 70,000 per day easily achieveable, depending on the CPU and o/c. There is still the hit on a running GPU app as always of course, however the new credit allocation makes it worthwhile now.

I've not run the AQUA ROQS app at the same time as the PPS Sieve GPU app, the latter takes three of my six cores on the 1090T cpu running 2x5970s, but the new higher AQUA credit allocation plus the relatively high credit on the GPU may make it worthwhile running both - not sure would need to test that.

Example Hosts running AQUA ROQS (ignore top two, they are currently being looked at as the results are suspiciously high, but you can see the effect of the new credit allocation put in place about a week ago (31,000 per WU - i7 920s / i7 940s do them in around 10-11hrs, which is the level expected from a 1090T).

Regards
Zy

Yes HassanShebli's 1090T took about 12.8 hours to complete a ROQ task, so about 57,600 per day.

BTW which driver did you use to run PrimeGrid PPS Sieve on 5970s without getting invalids on the second GPU core?

I thought OpenCL was still incompatible with HD 5970.

I thought OpenCL was still incompatible with HD 5970.

It should work, but the problem with OpenCL on it is it still can only only it on one its 2 GPUs

Yes, supporting only one core is what I meant by incompatible.
Hence my question to Zydor about driver/invalid results when using HD 5970 on PrimeGrid PPS Sieve.

Some people including myself have tried using a HD 5970 on PPS Sieve and not realised for a while that all work completed on the second core is marked invalid after it is sent to a wingman for validation. There is no indication at the time of processing or reporting tasks on the second core that anything is wrong except that these tasks go into pending and get sent out again to a wingman. When the wingman successfully completes and reports then these tasks are marked invalid. This may be a day or more later. Tasks completed on the first core validate quickly without being sent to a wingman.

Although I mentioned OpenCL being incompatible with HD 5970 some time ago on the PrimeGrid forum, there is still no general warning on the PrimeGrid website easily seen by new PrimeGrid contributors and every now and again a 5970 owner gets caught out. I assume this OpenCL incompatibility now also applies to HD 6990.

It was 11.3 (with APP). I didnt see any problems with gpu 2. I was running 2x5970 (CF), whether that made a difference, dont know.

I've got some 5oB's running at present for PG, witn Collatz on the GPUs. I will hold both for a while, and run PPS Sieve again, so you can see the result.

Murphy's Law states the dammed thing falls over now rofl ..... :)

They will be running in a few minutes, I'll keep it going for a few .

Regards
Zy

Got them running on the 2x5970s @840/500, usually do them @860+/500 but wont push it for this as its just a few to look and see.

I like the application, although there is a disadvantage in that it takes up 3 of my 6 cores on the 1090T (0.80 CPU per GPU, unless it changed recently)

Regards
Zy

I thought OpenCL was still incompatible with HD 5970.

It should work, but the problem with OpenCL on it is it still can only only it on one its 2 GPUs

When I run MilkyWay, with the 2x5970s, both cards are linked in Crossfire (running 1xWU per GPU - total four) - whether the later makes any difference to the OpenCL, not a clue :)

Regards
Zy

First four are through, 1980secs each, bit more than I was getting last time, but I'm not pushing the hardware as I am sitting in 27 degrees ambient indoors - like a greenhouse today.

Four WUs running concurrently on 2x5970s running 840/500. Hardware is 1090T @3.5Ghz, 16Gb RAM, 2x 1Tb hard disks. Computers not hidden if you want to look at the output details.

I'll keep the next four running to completion then switch back to the 5oB's etc - they are up against it time wise to complete the batch.

EDIT: Second batch of four PPS Sieve done, roughly the same time, should have been a min or so quicker as b uffers would have filled ..... but it was my fault, fiddling around (bit like Nero in Rome rofl) and I crashed it. Not the gpus fault :) Switching back to 5oB etc now.

Regards
Zy

Thanks, that's excellent. I read AMD were intending to fix it in SDK 2.4, but I didn't believe them because it has been so long now. I wonder if it works in all OpenCL applications or just PrimeGrid PPS Sieve. Thanks again, must give PPS Sieve a whirl myself now. :)

That one was one of Crunch3r's sterling efforts coming to the aid of us AMD users yet again ..... he might know, as I noticed today it has changed from V1.38 to V1.39, dont know if that was him following up, or someone else at PG.

Regards
Zy

Great, it's working without error on my 5970. I'm pleased to have another ATI GPU project to choose from. My Cat 11.4 had no APP so I had to install SDK 2.4.

Tasks are longer than before which is good. Previously the second core was a fair bit slower than the first but now they are the same. Tasks take about 36 minutes @ 770/500 so now about the same credit rate as here. Reasonably smooth GPU usage of about 94% with no stopping and starting every minute or two and power draw is much lower, main drawback as you pointed out is it uses a full CPU core per GPU task.

Will do a little more AQUA to compensate which brings me nicely back on topic. Had to reconfigure AQUA for 6 cores as it wouldn't start on previous 7 core configuration.