I never felt I really understood what you guys mean by 1x, 3x etc. The best I can come up with is the number of concurrent tasks. If that is so, then it seems counter-intuitive to me. How could running only 1 task at a time be more efficient than 3, if you have the resources to run 3?
<snip>
"Oh no," said the pot of petunias. "Not again!"
See my previous.
Quote:
your card will be much better suited to running only the O3AS tasks in my opinion. running BRP7 along side will slow them all down. i would set it up to be more like a prime/backup situation where you run BRP7 when O3AS is not available. but that's just my opinion, you can ultimately run it however you like.
your card will be much better suited to running only the O3AS tasks in my opinion. running BRP7 along side will slow them all down. i would set it up to be more like a prime/backup situation where you run BRP7 when O3AS is not available. but that's just my opinion, you can ultimately run it however you like.
1. select the appropriate venue/location that is for your host (likely 'Generic') in the top bold dropdown list titled "Preference Set"
2. under the "Applications" section unselect the BRP7 project titled "Binary Radio Pulsar Search (MeerKAT, GPU) [BRP7]" and leave O3AS selected
3. further down under the "Other Settings" section. change this option to YES:
Allow non-preferred apps: YES
4. save changes (bottom of page)
this will tell the project to send you only O3AS when they are available, and only send you other apps (which in this case includes BRP7) when O3AS is unavailable.
Thanks, but I already got all that long ago and it isn't even what I was asking.
I asked about the meaning of the jargon used in explaining certain things, like computing efficiency, not a discussion about that efficiency.
So again, when you are all speaking of, say, 5x, do you mean 5 concurrent tasks running?
To be fair, you did ask two questions. the meaning of the jargon, and the follow up asking how running 1x could be more efficient than 3x. that's what I attempted to answer.
Thanks, but I already got all that long ago and it isn't even what I was asking.
I asked about the meaning of the jargon used in explaining certain things, like computing efficiency, not a discussion about that efficiency.
So again, when you are all speaking of, say, 5x, do you mean 5 concurrent tasks running?
To be fair, you did ask two questions. the meaning of the jargon, and the follow up asking how running 1x could be more efficient than 3x. that's what I attempted to answer.
True, but at the time I wasn't able to follow what you were saying. It wasn't until I had my own GPU here that I could gather my own empirical data to understand what was going on.
I had at first assumed that a GPU and a CPU were pretty much similar, but the results showed one significant difference. The GPU is designed to process the data of a single process (PID), and one process only. When you ask it to process the data of 2 or more PIDs, things slow down significantly. I'm sure you said as much previously, but at the time I just didn't appreciate the significance.
So here I sit with a 16GB GPU when I only need 2. I can pick up a 6GB GeForce RTX 3050 for $500 less (taxes in) than this 4060 Ti cost me. Time for an RMA.
i would recommend something like a 3060Ti 8GB instead or 3070 depending on budget. much better memory setup.
the 6GB 3050 has a 96-bit memory bus. it will struggle.
The 8GB 3050 has a 128 bit bus, and is still affordable. I really cannot justify the price of the 4060 Ti I have here now, if all I can do with it is run single tasks. The 3060 Ti and 3070 all cost more than that, on both Amazon and Newegg.
KLiK wrote:Do you get it now?
)
Thanks, but I already got all that long ago and it isn't even what I was asking.
I asked about the meaning of the jargon used in explaining certain things, like computing efficiency, not a discussion about that efficiency.
So again, when you are all speaking of, say, 5x, do you mean 5 concurrent tasks running?
Ian&Steve C. wrote: hadron
)
"Oh no," said the pot of petunias. "Not again!"
See my previous.
I have no idea how I might do that.
hadron wrote: KLiK wrote:Do
)
Yes, I was talking about 5 concurrent tasks (CT)...as on basis of 24h (1 day) during more CT might be more efficient way to do so.
Not more efficient for your electric bill though! ;)
non-profit org. Play4Life in Zagreb, Croatia, EU
hadron wrote: Quote:your
)
go to your project settings here: https://einsteinathome.org/account/prefs/project
1. select the appropriate venue/location that is for your host (likely 'Generic') in the top bold dropdown list titled "Preference Set"
2. under the "Applications" section unselect the BRP7 project titled "Binary Radio Pulsar Search (MeerKAT, GPU) [BRP7]" and leave O3AS selected
3. further down under the "Other Settings" section. change this option to YES:
Allow non-preferred apps: YES
4. save changes (bottom of page)
this will tell the project to send you only O3AS when they are available, and only send you other apps (which in this case includes BRP7) when O3AS is unavailable.
_________________________________________________________________________
hadron wrote: Thanks, but I
)
To be fair, you did ask two questions. the meaning of the jargon, and the follow up asking how running 1x could be more efficient than 3x. that's what I attempted to answer.
_________________________________________________________________________
Ian&Steve C. wrote: hadron
)
True, but at the time I wasn't able to follow what you were saying. It wasn't until I had my own GPU here that I could gather my own empirical data to understand what was going on.
I had at first assumed that a GPU and a CPU were pretty much similar, but the results showed one significant difference. The GPU is designed to process the data of a single process (PID), and one process only. When you ask it to process the data of 2 or more PIDs, things slow down significantly. I'm sure you said as much previously, but at the time I just didn't appreciate the significance.
So here I sit with a 16GB GPU when I only need 2. I can pick up a 6GB GeForce RTX 3050 for $500 less (taxes in) than this 4060 Ti cost me. Time for an RMA.
i would recommend something
)
i would recommend something like a 3060Ti 8GB instead or 3070 depending on budget. much better memory setup.
the 6GB 3050 has a 96-bit memory bus. it will struggle.
_________________________________________________________________________
Ian&Steve C. wrote: i would
)
The 8GB 3050 has a 128 bit bus, and is still affordable. I really cannot justify the price of the 4060 Ti I have here now, if all I can do with it is run single tasks. The 3060 Ti and 3070 all cost more than that, on both Amazon and Newegg.
you can (and will benefit
)
you can (and will benefit from) run multiple tasks on the O3AS work. it's only the BRP7 tasks that will run best with 1x task.
_________________________________________________________________________
Ian&Steve C. wrote: you can
)
So on an 8 GB card, I can set <gpu_usage> for O3AS to 0.5 or even 0.33, while leaving the setting for BRP7 at 1?