Your RAC increase is not due to GW tasks - the new O3AS are taking longer for the same credit.
In the last couple of days, your machine has crunched a lot of GRP quite quickly. You have none left now but are seeing the result of the much higher credit award while they lasted.
For those still trying to figure out how to get the O3 tasks this is how:
1) Select one of the CPU-ONLY apps. Note that is CPU and not GPU. (I chose the first option, "Binary Radio Pulsar Search (Arecibo)".)
2) Select YES for the "Use Nvidia/AMD GPU" type and NO for "Use CPU". The combination of #1 and #2 effectively means you have nothing selected.
3) Select YES for "Allow non-preferred apps". This allows you to grab the 03ASE tasks which don't seem to yet have a selection box on the preferences screen.
Doing this, I got a bunch of the O3ASE GPU tasks and none of the other tasks.
Your RAC increase is not due to GW tasks - the new O3AS are taking longer for the same credit.
In the last couple of days, your machine has crunched a lot of GRP quite quickly. You have none left now but are seeing the result of the much higher credit award while they lasted.
I just upped my own crunching to 2 units at a time on my 6gb 1660Ti and it only creased the time by a few minutes which is very good compared to the 20 couple it was taking to do just one unit at a time. I did cut back on the number of cpu units I'm running to compensate so the gpu tasks each get the full allotment they want. I sure hope the validation happens soon or I could be trashing ALOT of tasks right now.
The most recent tasks are now taking about 35 minutes each so since the clear advantage is gone I'm going back to one at a time on this gpu.
Your RAC increase is not due to GW tasks - the new O3AS are taking longer for the same credit.
In the last couple of days, your machine has crunched a lot of GRP quite quickly. You have none left now but are seeing the result of the much higher credit award while they lasted.
Gary, thanks for thorough looking, I didnt do it this time.
For those still trying to figure out how to get the O3 tasks this is how:
1) Select one of the CPU-ONLY apps. Note that is CPU and not GPU. (I chose the first option, "Binary Radio Pulsar Search (Arecibo)".)
2) Select YES for the "Use Nvidia/AMD GPU" type and NO for "Use CPU". The combination of #1 and #2 effectively means you have nothing selected.
3) Select YES for "Allow non-preferred apps". This allows you to grab the 03ASE tasks which don't seem to yet have a selection box on the preferences screen.
Doing this, I got a bunch of the O3ASE GPU tasks and none of the other tasks.
Wonder where you got that from?
My 280x won't run these (need to start a thread about that) but the 1060 is doing fine.
They are piggies for RAM; take up to 1.5GB video RAM.
So 2 WU per every 3GB RAM is a decent starting point for optimization.
For those still trying to figure out how to get the O3 tasks this is how:
1) Select one of the CPU-ONLY apps. Note that is CPU and not GPU. (I chose the first option, "Binary Radio Pulsar Search (Arecibo)".)
2) Select YES for the "Use Nvidia/AMD GPU" type and NO for "Use CPU". The combination of #1 and #2 effectively means you have nothing selected.
3) Select YES for "Allow non-preferred apps". This allows you to grab the 03ASE tasks which don't seem to yet have a selection box on the preferences screen.
Doing this, I got a bunch of the O3ASE GPU tasks and none of the other tasks.
Wonder where you got that from?
My 280x won't run these (need to start a thread about that) but the 1060 is doing fine.
They are piggies for RAM; take up to 1.5GB video RAM.
So 2 WU per every 3GB RAM is a decent starting point for optimization.
I got it from a thread at wuprop where we discuss new apps from every project.
Did you make sure you have the correct Nivida or AMD setting set for the 2080?
I am running one unit at a time on my dedicated 1660Ti in a laptop and they are taking about 8 1/2 minutes for each task and using upto 100% of the gpu at times.
For those still trying to figure out how to get the O3 tasks this is how:
1) Select one of the CPU-ONLY apps. Note that is CPU and not GPU. (I chose the first option, "Binary Radio Pulsar Search (Arecibo)".)
2) Select YES for the "Use Nvidia/AMD GPU" type and NO for "Use CPU". The combination of #1 and #2 effectively means you have nothing selected.
3) Select YES for "Allow non-preferred apps". This allows you to grab the 03ASE tasks which don't seem to yet have a selection box on the preferences screen.
Doing this, I got a bunch of the O3ASE GPU tasks and none of the other tasks.
Wonder where you got that from?
My 280x won't run these (need to start a thread about that) but the 1060 is doing fine.
They are piggies for RAM; take up to 1.5GB video RAM.
So 2 WU per every 3GB RAM is a decent starting point for optimization.
I got it from a thread at wuprop where we discuss new apps from every project.
Did you make sure you have the correct Nivida or AMD setting set for the 2080?
I am running one unit at a time on my dedicated 1660Ti in a laptop and they are taking about 8 1/2 minutes for each task and using upto 100% of the gpu at times.
It's a Saphire Tri-X R9 280x and nothing I tried would get it to complete this WU (it sits at 0.5% usage)
280x's running SRBase TF now (it's 1:3 FP64 and should go back to Milkyway)
The downvolted (40% wattage) 1060 is doing this and seems to be using 100% CPU and 85% GPU and completing in 38 minutes.
This WU requires OpenCL and so needs NVidia drivers 465+ (pushing video drivers 3 years past the cards' manufacture date usually causes issues because manufacturers want you to buy newer cards and obsolete cards with driver modifications to cause the card to malfunction, but it helped in this case).
Too hot outside, now, to run the GPU's full power.
It's a Saphire Tri-X R9 280x and nothing I tried would get it to complete this WU (it sits at 0.5% usage)
280x's running SRBase TF now (it's 1:3 FP64 and should go back to Milkyway)
The downvolted (40% wattage) 1060 is doing this and seems to be using 100% CPU and 85% GPU and completing in 38 minutes.
This WU requires OpenCL and so needs NVidia drivers 465+ (pushing video drivers 3 years past the cards' manufacture date usually causes issues because manufacturers want you to buy newer cards and obsolete cards with driver modifications to cause the card to malfunction, but it helped in this case).
Too hot outside, now, to run the GPU's full power.
I thought so I'm mikey on every project.
I did not know they required driver ver 465 or above but I do happen to be using ver 465.
I have no clue why the 2080 won't finish the units that is strange. I'm assuming you read the forum at WCG and found people who are running them with 2080's just fine.
I run the newer driver versions on my older 1080Ti cards, when they are running, and since I don't game they seem to work for me for crunching. Now I almost never jump on the latest and greatest driver but I'm usually only a few versions behind, just in case they crunch faster.
Your RAC increase is not due
)
Your RAC increase is not due to GW tasks - the new O3AS are taking longer for the same credit.
In the last couple of days, your machine has crunched a lot of GRP quite quickly. You have none left now but are seeing the result of the much higher credit award while they lasted.
Cheers,
Gary.
For those still trying to
)
For those still trying to figure out how to get the O3 tasks this is how:
1) Select one of the CPU-ONLY apps. Note that is CPU and not GPU. (I chose the first option, "Binary Radio Pulsar Search (Arecibo)".)
2) Select YES for the "Use Nvidia/AMD GPU" type and NO for "Use CPU". The combination of #1 and #2 effectively means you have nothing selected.
3) Select YES for "Allow non-preferred apps". This allows you to grab the 03ASE tasks which don't seem to yet have a selection box on the preferences screen.
Doing this, I got a bunch of the O3ASE GPU tasks and none of the other tasks.
Gary Roberts wrote:Your RAC
)
I just upped my own crunching to 2 units at a time on my 6gb 1660Ti and it only creased the time by a few minutes which is very good compared to the 20 couple it was taking to do just one unit at a time. I did cut back on the number of cpu units I'm running to compensate so the gpu tasks each get the full allotment they want. I sure hope the validation happens soon or I could be trashing ALOT of tasks right now.
The most recent tasks are now taking about 35 minutes each so since the clear advantage is gone I'm going back to one at a time on this gpu.
Gary Roberts wrote: Your RAC
)
Gary, thanks for thorough looking, I didnt do it this time.
mikey wrote: For those still
)
Wonder where you got that from?
My 280x won't run these (need to start a thread about that) but the 1060 is doing fine.
They are piggies for RAM; take up to 1.5GB video RAM.
So 2 WU per every 3GB RAM is a decent starting point for optimization.
marmot wrote: mikey
)
I got it from a thread at wuprop where we discuss new apps from every project.
Did you make sure you have the correct Nivida or AMD setting set for the 2080?
I am running one unit at a time on my dedicated 1660Ti in a laptop and they are taking about 8 1/2 minutes for each task and using upto 100% of the gpu at times.
mikey wrote: marmot
)
I'm 'marmot' on all BOINC projects except WCG (marm0t) and BAM(can't use the account manager).
This is me at WUProps about this WU: http://wuprop.boinc-af.org/forum_thread.php?id=191&postid=8097#8097
It's a Saphire Tri-X R9 280x and nothing I tried would get it to complete this WU (it sits at 0.5% usage)
280x's running SRBase TF now (it's 1:3 FP64 and should go back to Milkyway)
The downvolted (40% wattage) 1060 is doing this and seems to be using 100% CPU and 85% GPU and completing in 38 minutes.
This WU requires OpenCL and so needs NVidia drivers 465+ (pushing video drivers 3 years past the cards' manufacture date usually causes issues because manufacturers want you to buy newer cards and obsolete cards with driver modifications to cause the card to malfunction, but it helped in this case).
Too hot outside, now, to run the GPU's full power.
marmot wrote: I'm 'marmot'
)
I thought so I'm mikey on every project.
I did not know they required driver ver 465 or above but I do happen to be using ver 465.
I have no clue why the 2080 won't finish the units that is strange. I'm assuming you read the forum at WCG and found people who are running them with 2080's just fine.
I run the newer driver versions on my older 1080Ti cards, when they are running, and since I don't game they seem to work for me for crunching. Now I almost never jump on the latest and greatest driver but I'm usually only a few versions behind, just in case they crunch faster.