Join PrimeGrid
Returning Participants
Community
Leader Boards
Results
Other
drummers-lowrise
|
Message boards :
Prime Sierpinski Problem :
Optimizing PSP processing with LLR2 ....
| Author |
Message |
|---|
|
|
|
A question for the group: how should I allocate the six physical cores of my i7 8700k CPU when processing PSP units? Do I go after 3x2 cores, 2x3 cores or perhaps 1x4 / 1x5 / 1x6 cores?
I am running a task right now with 5 cores allocated to it on my i7 8700k and it seems that it will take about 27 hrs to complete. Similarly, another system (i3 6300) will require about 32 hrs with only two cores to complete, thus suggesting that the five cores with the i7 8700k are an overkill and I am wasting potential resources.
With LLR2 it is not vital that we rush to return a task and, as a result, I can easily afford a slightly slower return if it will mean greater throughput per unit of time.
Thanks! | |
|
 Nick 
Send message
Joined: 11 Jul 11
Posts: 1718
ID: 105020
Credit: 4,338,128,139
RAC: 10,463,363
                        
|
|
I can answer this quickly.
Do 1 task with all cores.
Edit: I am now paying attention. You will not get better performance on these CPUs than 1 task running. | |
|
|
|
|
Thanks, Nick! | |
|
RafaelVolunteer tester
Send message
Joined: 22 Oct 14
Posts: 905
ID: 370496
Credit: 459,496,650
RAC: 148,364
                  
|
A question for the group: how should I allocate the six physical cores of my i7 8700k CPU when processing PSP units? Do I go after 3x2 cores, 2x3 cores or perhaps 1x4 / 1x5 / 1x6 cores?
I am running a task right now with 5 cores allocated to it on my i7 8700k and it seems that it will take about 27 hrs to complete. Similarly, another system (i3 6300) will require about 32 hrs with only two cores to complete, thus suggesting that the five cores with the i7 8700k are an overkill and I am wasting potential resources.
With LLR2 it is not vital that we rush to return a task and, as a result, I can easily afford a slightly slower return if it will mean greater throughput per unit of time.
Thanks!
Rule of Thumb #1: always have the sum of cache usage for all your tasks a bit below your CPU cache.
Looking at single one of your tasks, it's using 2880k * 8 = ~23MB of cache, which is above your CPU's 12MB limit. So you should run fewer tasks to reduce usage... but since you can't run fewer than one, just run that single task with as many cores as you have. | |
|
Nick
Send message
Joined: 11 Jul 11
Posts: 1718
ID: 105020
Credit: 4,338,128,139
RAC: 10,463,363
|
Rule of Thumb #1: always have the sum of cache usage for all your tasks a bit below your CPU cache..
Except it becomes 'weird' with the Intel CPUs that don't mirror information between L2 and L3.
I could have sworn that running 2 x PSP was not the best best choice, but I saw what Pooh Bear (Thankyou) was doing on his 10980XE and it worked for my 9960Xs too.
This isn't in contradiction of what you are saying.
Keep it below cache capacity, if you can. | |
|
|
|
Rule of Thumb #1: always have the sum of cache usage for all your tasks a bit below your CPU cache.
Looking at single one of your tasks, it's using 2880k * 8 = ~23MB of cache, which is above your CPU's 12MB limit. So you should run fewer tasks to reduce usage... but since you can't run fewer than one, just run that single task with as many cores as you have.
Many thanks for the reminder, Rafale!
I have another unit completing in 90 minutes and which will come in at about 3 hrs earlier than expected and I suspect it will be due to a difference in FFT length (ex 2340k vs 2880k). | |
|
BurVolunteer tester
Send message
Joined: 25 Feb 20
Posts: 511
ID: 1241833
Credit: 406,968,745
RAC: 28,964
               
|
|
You can also use Prime95/Mprime benchmark feature that tests the throughput for different cores/LLR combinations depending on the FFT size. Just make sure to enable "all-complex FFT".
That'll give you a very good estimate which number of cores is optimal.
____________
1281979 * 2^485014 + 1 is prime ... no further hits up to: n = 5,700,000 | |
|
|
|
|
how do I know what my cache is?
____________
Thanks,
Jim
 | |
|
|
|
how do I know what my cache is?
I don't know an app nor a command for Linux.
Anyway, I believe the info can be found in the chip manufacturer website.
For Windows there is CPU-Z.
____________
"Accidit in puncto, quod non contingit in anno."
Something that does not occur in a year may, perchance, happen in a moment. | |
|
|
|
how do I know what my cache is?
Your i5-5250U: 3MB
Your i5-10600: 12MB
(For everybody) Wikipedia is also a great place to look up data on whole families of chips, which is great comparison material for those shopping for new or used hardware, particularly the still excellent and cheap retired server stuff.
____________
Eating more cheese on Thursdays. | |
|
|
|
|
Thanks...
Jim | |
|
streamVolunteer moderator
Project administrator
Volunteer developer
Volunteer tester
Send message
Joined: 1 Mar 14
Posts: 981
ID: 301928
Credit: 543,185,506
RAC: 36,711
           
|
how do I know what my cache is?
I don't know an app nor a command for Linux.
In Linux, everything is in /proc/cpuinfo - just read this file or filter necessary lines:
cat /proc/cpuinfo | grep cache
| |
|
Post to thread
Message boards :
Prime Sierpinski Problem :
Optimizing PSP processing with LLR2 .... |
