PrimeGrid
Please visit donation page to help the project cover running costs for this month

Toggle Menu

Join PrimeGrid

Returning Participants

Community

Leader Boards

Results

Other

drummers-lowrise

Advanced search

Message boards : Number crunching : Automatic subproject recommendation for each host

Author Message
Profile composite
Volunteer tester
Send message
Joined: 16 Feb 10
Posts: 764
ID: 55391
Credit: 688,170,015
RAC: 74,625
Discovered 2 mega primesFound 1 prime in the 2018 Tour de Primes321 LLR Turquoise: Earned 5,000,000 credits (5,477,467)Cullen LLR Gold: Earned 500,000 credits (776,297)ESP LLR Ruby: Earned 2,000,000 credits (3,120,351)Generalized Cullen/Woodall LLR Ruby: Earned 2,000,000 credits (2,056,207)PPS LLR Jade: Earned 10,000,000 credits (18,272,027)PSP LLR Turquoise: Earned 5,000,000 credits (5,027,818)SoB LLR Sapphire: Earned 20,000,000 credits (24,671,131)SR5 LLR Turquoise: Earned 5,000,000 credits (6,110,877)SGS LLR Ruby: Earned 2,000,000 credits (3,477,744)TRP LLR Turquoise: Earned 5,000,000 credits (7,025,303)Woodall LLR Amethyst: Earned 1,000,000 credits (1,693,614)321 Sieve Emerald: Earned 50,000,000 credits (50,256,050)Cullen/Woodall Sieve (suspended) Turquoise: Earned 5,000,000 credits (5,571,178)Generalized Cullen/Woodall Sieve (suspended) Emerald: Earned 50,000,000 credits (50,009,610)PPS Sieve Double Silver: Earned 200,000,000 credits (312,256,351)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Jade: Earned 10,000,000 credits (10,165,888)TRP Sieve (suspended) Sapphire: Earned 20,000,000 credits (20,071,454)AP 26/27 Turquoise: Earned 5,000,000 credits (6,616,128)GFN Emerald: Earned 50,000,000 credits (52,755,348)PSA Double Bronze: Earned 100,000,000 credits (102,762,384)
Message 142417 - Posted: 13 Aug 2020 | 3:20:16 UTC

This is a feature request.

I sure would like to see PG offer recommendations for subprojects for each of my hosts according to its CPU, and perhaps also according my specific objectives (say, CPU efficiency or rate of credit production). I am particularly interested in CPU efficiency, so I will focus on this.

It's labour-intensive and time-consuming to manually collect task statistics for each of my hosts across the spectrum of active subprojects, so it's hard to figure out which subprojects are best suited for each host. The "fastest CPU" list is not specific enough for the performance profile of each CPU on subprojects.

But PG has a database of tasks and CPUs, so it's feasible for the server to compute the median CPU time of valid tasks for each subproject by CPU type across all hosts in the database and present a list ranked by CPU time for each host.

It's only a recommendation since each individual host has a distinctive hardware and software configuration that makes it unlikely to perform exactly like the median statistic. However, it's better than nothing. It's also better than the statistical average because the average is affected by the subset of hosts using multitasking options.

The sampling can be periodic so that it's low intensity on the server, and once the median statistic has been computed from a sufficiently large population of tasks for a particular CPU type and subproject, that statistic never needs to be computed again. The statistic is stored for reference by the recommendation system.

Profile j.sheridanProject donor
Send message
Joined: 21 Mar 11
Posts: 659
ID: 91622
Credit: 1,108,930,436
RAC: 2,168,048
321 LLR Turquoise: Earned 5,000,000 credits (5,019,701)Cullen LLR Ruby: Earned 2,000,000 credits (4,582,271)ESP LLR Ruby: Earned 2,000,000 credits (4,563,800)Generalized Cullen/Woodall LLR Ruby: Earned 2,000,000 credits (3,975,172)PPS LLR Turquoise: Earned 5,000,000 credits (8,639,452)PSP LLR Turquoise: Earned 5,000,000 credits (5,500,951)SoB LLR Turquoise: Earned 5,000,000 credits (8,583,226)SR5 LLR Ruby: Earned 2,000,000 credits (3,441,728)SGS LLR Turquoise: Earned 5,000,000 credits (5,169,579)TRP LLR Turquoise: Earned 5,000,000 credits (9,717,538)Woodall LLR Ruby: Earned 2,000,000 credits (4,084,622)321 Sieve Sapphire: Earned 20,000,000 credits (20,019,388)Cullen/Woodall Sieve (suspended) Double Silver: Earned 200,000,000 credits (265,102,350)PPS Sieve Double Gold: Earned 500,000,000 credits (514,289,554)TRP Sieve (suspended) Turquoise: Earned 5,000,000 credits (5,801,812)AP 26/27 Emerald: Earned 50,000,000 credits (53,472,718)GFN Double Bronze: Earned 100,000,000 credits (186,978,187)
Message 142420 - Posted: 13 Aug 2020 | 7:33:45 UTC - in response to Message 142417.

It's a good idea but the problem I see is that, for whatever reason, people don't bother investigating the optimal configuration for their CPU.
You'd need to see a list of tasks and the number of threads used to run that task.
At that point you'd still need to work out the right number of threads to use for your processor for that subproject which just to complicate matters can actually vary depending on the task.
As an example at the extreme end, I can run SoB on a 3700x (faster CPUs are available) in around 70000 seconds.
I've seen from my wingmen that plenty of people just run single threaded. If you tried to run 8 SoB tasks on a 3700x it would be closer to 1000000 seconds per task. So now for the same processor and the same task you have times of 70000 and 1000000 and everything in between.
What do you do?

Profile DaveProject donor
Avatar
Send message
Joined: 13 Feb 12
Posts: 2781
ID: 130544
Credit: 934,927,378
RAC: 374,340
Found 2 primes in the 2018 Tour de PrimesFound 1 prime in the 2020 Tour de Primes321 LLR Turquoise: Earned 5,000,000 credits (6,042,688)Cullen LLR Turquoise: Earned 5,000,000 credits (5,101,836)ESP LLR Turquoise: Earned 5,000,000 credits (5,303,526)Generalized Cullen/Woodall LLR Turquoise: Earned 5,000,000 credits (5,202,873)PPS LLR Turquoise: Earned 5,000,000 credits (6,100,156)PSP LLR Turquoise: Earned 5,000,000 credits (5,513,728)SoB LLR Turquoise: Earned 5,000,000 credits (8,851,996)SR5 LLR Turquoise: Earned 5,000,000 credits (5,199,189)SGS LLR Turquoise: Earned 5,000,000 credits (5,100,018)TRP LLR Turquoise: Earned 5,000,000 credits (5,405,027)Woodall LLR Turquoise: Earned 5,000,000 credits (5,124,820)321 Sieve Jade: Earned 10,000,000 credits (10,003,334)Cullen/Woodall Sieve (suspended) Silver: Earned 100,000 credits (268,250)Generalized Cullen/Woodall Sieve (suspended) Jade: Earned 10,000,000 credits (10,000,502)PPS Sieve Double Silver: Earned 200,000,000 credits (306,761,000)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Jade: Earned 10,000,000 credits (10,000,133)TRP Sieve (suspended) Jade: Earned 10,000,000 credits (10,000,970)AP 26/27 Double Bronze: Earned 100,000,000 credits (191,622,028)GFN Double Bronze: Earned 100,000,000 credits (133,333,389)PSA Double Silver: Earned 200,000,000 credits (200,000,001)
Message 142421 - Posted: 13 Aug 2020 | 8:45:57 UTC

The principle is good. In reality there may be too many edge cases to make it a reliable option.

Profile dannyridel
Volunteer tester
Avatar
Send message
Joined: 3 Feb 19
Posts: 739
ID: 1097922
Credit: 2,595,126
RAC: 4,929
321 LLR Bronze: Earned 10,000 credits (78,095)Cullen LLR Bronze: Earned 10,000 credits (36,962)ESP LLR Bronze: Earned 10,000 credits (36,309)Generalized Cullen/Woodall LLR Bronze: Earned 10,000 credits (22,373)PPS LLR Silver: Earned 100,000 credits (177,303)PSP LLR Bronze: Earned 10,000 credits (65,677)SoB LLR Bronze: Earned 10,000 credits (66,029)SR5 LLR Bronze: Earned 10,000 credits (53,645)SGS LLR Bronze: Earned 10,000 credits (22,711)TRP LLR Bronze: Earned 10,000 credits (67,784)Woodall LLR Bronze: Earned 10,000 credits (19,031)321 Sieve Gold: Earned 500,000 credits (506,814)Generalized Cullen/Woodall Sieve (suspended) Bronze: Earned 10,000 credits (30,033)PPS Sieve Gold: Earned 500,000 credits (714,652)AP 26/27 Bronze: Earned 10,000 credits (80,860)GFN Silver: Earned 100,000 credits (243,814)PSA Silver: Earned 100,000 credits (373,034)
Message 142423 - Posted: 13 Aug 2020 | 10:29:31 UTC - in response to Message 142417.

I agree with dave.

BUT I still think there to be a plausible option:
Choose only between the shorter tasks, etc. PPSE, SGS, PPS, MEGA, DIV, SR5, 321 SV.


I've seen from my wingmen that plenty of people just run single threaded. If you tried to run 8 SoB tasks on a 3700x it would be closer to 1000000 seconds per task. So now for the same processor and the same task you have times of 70000 and 1000000 and everything in between.
What do you do?


Wellll
My Ryzen5 2500U (zen1) runs SoB 4-threaded in 1.4m seconds. And most ppl who don't look at their settings have CPUs that perform like mine. So this further justifies my previous suggestion of only bringing up the shorter tasks.
____________
SHSID Electronics Group
SHSIDElectronicsGroup@outlook.com

GFN-14: 50103906^16384+1
Proth "SoB": 44243*2^440969+1


Profile mikey
Avatar
Send message
Joined: 17 Mar 09
Posts: 1191
ID: 37043
Credit: 505,311,599
RAC: 131,529
Discovered 1 mega prime321 LLR Ruby: Earned 2,000,000 credits (2,038,739)Cullen LLR Ruby: Earned 2,000,000 credits (2,074,615)ESP LLR Ruby: Earned 2,000,000 credits (2,013,823)Generalized Cullen/Woodall LLR Ruby: Earned 2,000,000 credits (2,142,353)PPS LLR Turquoise: Earned 5,000,000 credits (5,149,318)PSP LLR Ruby: Earned 2,000,000 credits (2,049,284)SoB LLR Amethyst: Earned 1,000,000 credits (1,043,824)SR5 LLR Ruby: Earned 2,000,000 credits (2,053,250)SGS LLR Turquoise: Earned 5,000,000 credits (5,095,282)TRP LLR Ruby: Earned 2,000,000 credits (2,025,737)Woodall LLR Amethyst: Earned 1,000,000 credits (1,893,227)321 Sieve Sapphire: Earned 20,000,000 credits (23,770,672)Cullen/Woodall Sieve (suspended) Gold: Earned 500,000 credits (944,431)Generalized Cullen/Woodall Sieve (suspended) Sapphire: Earned 20,000,000 credits (20,813,253)PPS Sieve Double Silver: Earned 200,000,000 credits (339,648,557)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Ruby: Earned 2,000,000 credits (2,446,797)AP 26/27 Sapphire: Earned 20,000,000 credits (33,140,471)GFN Sapphire: Earned 20,000,000 credits (36,517,558)PSA Sapphire: Earned 20,000,000 credits (20,457,430)
Message 142425 - Posted: 13 Aug 2020 | 11:15:06 UTC - in response to Message 142420.

It's a good idea but the problem I see is that, for whatever reason, people don't bother investigating the optimal configuration for their CPU.
You'd need to see a list of tasks and the number of threads used to run that task.
At that point you'd still need to work out the right number of threads to use for your processor for that subproject which just to complicate matters can actually vary depending on the task.
As an example at the extreme end, I can run SoB on a 3700x (faster CPUs are available) in around 70000 seconds.
I've seen from my wingmen that plenty of people just run single threaded. If you tried to run 8 SoB tasks on a 3700x it would be closer to 1000000 seconds per task. So now for the same processor and the same task you have times of 70000 and 1000000 and everything in between.
What do you do?


One option would be to show both single threaded and multi-threaded times, as the OP says PG has hundreds of thousands of hours of stats for each model cpu and sub-project. Sure not everyone runs things optimally but a number is better than no number at this point where nothing is known. I don't want to take away from the OP's idea but if PG would just pick a cpu type and run the numbers for a single type of task to see how hard it would be and what kind of numbers come up maybe the OP, and others, could help make the data usable by people who are ALWAYS asking...'what and how should I run it on my cpu?'

Sure lots of people run projects for badges but some people would prefer to run things that just work well on their machines too. It would also help answer the questions...'how is my machine doing compared to other machines running the same task with the same cpu'? In the end it may even help those of us waiting for that wingman running the million second task when we finished it in 70 thousand seconds.

mackerelProject donor
Volunteer tester
Avatar
Send message
Joined: 2 Oct 08
Posts: 2420
ID: 29980
Credit: 417,913,246
RAC: 165,476
Discovered 2 mega primesEliminated 1 conjecture "k"Found 3 primes in the 2018 Tour de PrimesFound 1 mega prime in the 2018 Tour de PrimesFound 5 primes in the 2019 Tour de PrimesFound 6 primes in the 2020 Tour de Primes321 LLR Turquoise: Earned 5,000,000 credits (8,774,878)Cullen LLR Turquoise: Earned 5,000,000 credits (5,149,818)ESP LLR Turquoise: Earned 5,000,000 credits (6,454,573)Generalized Cullen/Woodall LLR Turquoise: Earned 5,000,000 credits (5,122,074)PPS LLR Emerald: Earned 50,000,000 credits (76,916,935)PSP LLR Jade: Earned 10,000,000 credits (15,223,714)SoB LLR Jade: Earned 10,000,000 credits (17,319,914)SR5 LLR Sapphire: Earned 20,000,000 credits (23,996,561)SGS LLR Turquoise: Earned 5,000,000 credits (7,342,780)TPS LLR (retired) Bronze: Earned 10,000 credits (34,130)TRP LLR Jade: Earned 10,000,000 credits (18,602,519)Woodall LLR Turquoise: Earned 5,000,000 credits (5,715,464)321 Sieve Sapphire: Earned 20,000,000 credits (20,236,219)Cullen/Woodall Sieve (suspended) Turquoise: Earned 5,000,000 credits (5,383,853)Generalized Cullen/Woodall Sieve (suspended) Sapphire: Earned 20,000,000 credits (20,626,419)PPS Sieve Emerald: Earned 50,000,000 credits (76,969,144)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Ruby: Earned 2,000,000 credits (2,293,882)TRP Sieve (suspended) Turquoise: Earned 5,000,000 credits (5,012,757)AP 26/27 Sapphire: Earned 20,000,000 credits (21,918,894)GFN Emerald: Earned 50,000,000 credits (71,887,807)PSA Ruby: Earned 2,000,000 credits (2,939,755)
Message 142426 - Posted: 13 Aug 2020 | 12:03:25 UTC

I don't know about optimal subprojects, but as a subset of that, would be optimising the running of LLR work beyond the default 1 task per thread. Knowing the CPU model it is not too difficult to choose good settings for LLR throughput. Not necessarily best, but no worse than no configuration at all. The difficulty is having a database of CPUs in order to choose, and working within the limits of any user customisation to how BOINC is running.

If consumer Intel Core CPU since Sandy Bridge with AVX enabled (may exclude Celeron and Pentium models): look up L3 cache. Pick number of simultaneous tasks such that max_FFT_size*8*num_tasks does not exceed L3 cache. In case of doubt, pick fewer simultaneous tasks. Set number of threads per task accordingly. If HT is on and BOINC CPU thread count is not reduced below max, it is better to waste power on using the HT in these tasks, than risk them starting additional tasks. Unless it is possible to lie to BOINC. e.g. say the tasks uses 8 thread, but only actually use 4. This may raise support issues with people saying why isn't my CPU at 100% load?

If Skylake-X or server equivalents, same as above but use total L2+L3.

If Zen 2, same as above, but with the additional constraint task(s) should not cross a CCX unless absolutely unavoidable. e.g. for a 6 core CPU (2x 3-core CCX), do not run 3 x 2-core tasks, jump to 2x3c if 6x1c is not longer optimal. This implies that 4 cores will be the limit for most tasks, only the biggest tasks might run outside that.

If Zen/Zen+ or older, Intel without AVX: above should still work, but I'd rather put them on sieve since that isn't impacted from a weak FPU.

I'm sure there are many edge cases not covered where it is ambiguous what to choose. I have no idea how AP27 or GFN fits into this. I guess GFN is more GPU optimal these days.

Scott BrownProject donor
Volunteer moderator
Project administrator
Volunteer tester
Project scientist
Avatar
Send message
Joined: 17 Oct 05
Posts: 2125
ID: 1178
Credit: 8,410,720,955
RAC: 6,716,484
Discovered the World's First base 116 Generalized Cullen prime!!!Discovered 26 mega primesEliminated 7 conjecture "k"sDiscovered 1 Sophie Germain pairDiscovered 2 Fermat divisors2012 Tour de Primes highest prime count2012 Tour de Primes most Mountain Stage primes2015 Tour de Primes highest prime count2016 Tour de Primes highest prime countFound 23 primes in the 2018 Tour de PrimesFound 1 mega prime in the 2018 Tour de PrimesFound 2 primes in the 2018 Tour de Primes Mountain Stage2019 Tour de Primes highest prime countFound 22 primes in the 2019 Tour de Primes2020 Tour de Primes highest prime scoreFound 21 primes in the 2020 Tour de PrimesFound 4 mega primes in the 2020 Tour de Primes321 LLR Double Bronze: Earned 100,000,000 credits (185,327,043)Cullen LLR Double Bronze: Earned 100,000,000 credits (103,870,990)ESP LLR Double Silver: Earned 200,000,000 credits (203,249,784)Generalized Cullen/Woodall LLR Double Bronze: Earned 100,000,000 credits (109,580,172)PPS LLR Double Gold: Earned 500,000,000 credits (639,557,324)PSP LLR Double Bronze: Earned 100,000,000 credits (126,982,721)SoB LLR Double Bronze: Earned 100,000,000 credits (135,747,083)SR5 LLR Double Silver: Earned 200,000,000 credits (214,194,272)SGS LLR Double Silver: Earned 200,000,000 credits (200,486,586)TPS LLR (retired) Silver: Earned 100,000 credits (235,439)TRP LLR Double Silver: Earned 200,000,000 credits (201,215,056)Woodall LLR Double Bronze: Earned 100,000,000 credits (101,447,725)321 Sieve Double Silver: Earned 200,000,000 credits (235,451,253)Cullen/Woodall Sieve (suspended) Emerald: Earned 50,000,000 credits (83,794,448)Generalized Cullen/Woodall Sieve (suspended) Double Silver: Earned 200,000,000 credits (285,139,652)PPS Sieve Double Ruby: Earned 2,000,000,000 credits (2,665,703,405)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Double Silver: Earned 200,000,000 credits (203,523,358)TRP Sieve (suspended) Double Silver: Earned 200,000,000 credits (201,489,157)AP 26/27 Double Silver: Earned 200,000,000 credits (374,978,656)GFN Double Amethyst: Earned 1,000,000,000 credits (1,879,778,210)PSA Double Silver: Earned 200,000,000 credits (259,058,048)
Message 142427 - Posted: 13 Aug 2020 | 12:07:41 UTC

While I appreciate the reasoning behind this request, it is difficult to see how this would be feasible. More importantly, it is very unlikely to provide the information that is being sought because optimal (or anything approaching optimal) configuration suggestions are driven by a substantially large number of variables, many of which cannot be tracked by PG or the BOINC system.

Some of the obvious variables have been noted in the posts in this thread such as hardware configurations and differential software. Ideally, these could be tracked in an informative manner (e.g., hardware, OS, driver versions, etc.), but the reality is that the BOINC system does a very bad job of this quite often.

An easy example of this is how a GPU is detailed within BOINC. On systems with more than one GPU, BOINC records the GPU models as being both the same model--whichever GPU is device 0. For example, if I had a system with an RTX 2070 as device 0 and a GTX 750Ti as device 1, both would be recorded as RTX 2070s by BOINC, and the performance statistics would be artificially slowing the average (or medians or any other measure) for RTX 2070 cards. If I switched the device locations in my physical machine, then the performance of GTX 750Ti cards would be artificially enhanced in BOINC data. There is no way to correct for this in the BOINC system, and while it may be surprising to some, such mixed card model systems are actually quite common. This is not limited to GPUs as there are ways in which multiple systems can be made to report their results through a connecting system such that a single CPU system can look to have ridiculously fast and numerous results.

And that is just the hardware side of things. Even assuming that we could accurately see the specifics of hardware and software for every machine in the database, we would still have unavailable variables that affect optimal performance numbers tremendously. These include things like ambient conditions of a machine (e.g., heat can cause drastic performance differences in Pascal and Turing GPUs, etc.) and individual usage issues (other software being used on the machine can cause substantial performance variation on otherwise identical machines) to name but a couple.

In short, while this would be a nice feature, it would be terribly difficult to implement anything even approaching a useful set of recommendations, and those recommendations would far too often lack any greater accuracy than some of the general guidelines currently known and posted around the forum (e.g., which projects work well or not with multi-threading, how cache helps with larger tasks showing how FFT size corresponds to needed cache size, etc.).



Profile dannyridel
Volunteer tester
Avatar
Send message
Joined: 3 Feb 19
Posts: 739
ID: 1097922
Credit: 2,595,126
RAC: 4,929
321 LLR Bronze: Earned 10,000 credits (78,095)Cullen LLR Bronze: Earned 10,000 credits (36,962)ESP LLR Bronze: Earned 10,000 credits (36,309)Generalized Cullen/Woodall LLR Bronze: Earned 10,000 credits (22,373)PPS LLR Silver: Earned 100,000 credits (177,303)PSP LLR Bronze: Earned 10,000 credits (65,677)SoB LLR Bronze: Earned 10,000 credits (66,029)SR5 LLR Bronze: Earned 10,000 credits (53,645)SGS LLR Bronze: Earned 10,000 credits (22,711)TRP LLR Bronze: Earned 10,000 credits (67,784)Woodall LLR Bronze: Earned 10,000 credits (19,031)321 Sieve Gold: Earned 500,000 credits (506,814)Generalized Cullen/Woodall Sieve (suspended) Bronze: Earned 10,000 credits (30,033)PPS Sieve Gold: Earned 500,000 credits (714,652)AP 26/27 Bronze: Earned 10,000 credits (80,860)GFN Silver: Earned 100,000 credits (243,814)PSA Silver: Earned 100,000 credits (373,034)
Message 142429 - Posted: 13 Aug 2020 | 14:57:25 UTC

To add to the above posts, I personally think 321 sieve is great as a starter project.

Also, another very rare side case would be advanced coders like Pavel. Send 'optimized' MEGA tasks to a bunch of cache-full Xeons? I would not do so.
____________
SHSID Electronics Group
SHSIDElectronicsGroup@outlook.com

GFN-14: 50103906^16384+1
Proth "SoB": 44243*2^440969+1


Profile Michael GoetzProject donor
Volunteer moderator
Project administrator
Avatar
Send message
Joined: 21 Jan 10
Posts: 13400
ID: 53948
Credit: 229,156,034
RAC: 138,597
The "Shut up already!" badge:  This loud mouth has mansplained on the forums over 10 thousand times!  Sheesh!!!Discovered the World's First GFN-19 prime!!!Discovered 1 mega primeFound 1 prime in the 2018 Tour de PrimesFound 1 prime in the 2019 Tour de PrimesFound 1 prime in the 2020 Tour de Primes321 LLR Ruby: Earned 2,000,000 credits (2,822,730)Cullen LLR Ruby: Earned 2,000,000 credits (2,005,249)ESP LLR Turquoise: Earned 5,000,000 credits (5,009,577)Generalized Cullen/Woodall LLR Ruby: Earned 2,000,000 credits (2,145,754)PPS LLR Jade: Earned 10,000,000 credits (15,528,495)PSP LLR Turquoise: Earned 5,000,000 credits (5,197,957)SoB LLR Sapphire: Earned 20,000,000 credits (34,291,181)SR5 LLR Jade: Earned 10,000,000 credits (10,007,110)SGS LLR Ruby: Earned 2,000,000 credits (2,276,330)TRP LLR Turquoise: Earned 5,000,000 credits (5,084,329)Woodall LLR Ruby: Earned 2,000,000 credits (2,195,123)321 Sieve Jade: Earned 10,000,000 credits (10,061,196)Cullen/Woodall Sieve (suspended) Ruby: Earned 2,000,000 credits (4,170,256)Generalized Cullen/Woodall Sieve (suspended) Turquoise: Earned 5,000,000 credits (5,059,304)PPS Sieve Sapphire: Earned 20,000,000 credits (22,885,121)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Amethyst: Earned 1,000,000 credits (1,035,522)TRP Sieve (suspended) Ruby: Earned 2,000,000 credits (2,051,121)AP 26/27 Jade: Earned 10,000,000 credits (10,118,303)GFN Emerald: Earned 50,000,000 credits (74,768,242)PSA Jade: Earned 10,000,000 credits (12,445,029)
Message 142430 - Posted: 13 Aug 2020 | 15:06:02 UTC

Another problem is that any such system would think that CPUs lacking hyperthreading (typically i5) are more powerful than CPUs with hyperthreading.

Some percentage of hyperthreaded CPUs will run LLR with HT on, badly and irreversibly corrupting their performance data.

I agree with Scott, although my answer would have been shorter. Probably just one word. :)

It would be a horrendously difficult system to build and a huge effort would be required to maintain it moving forward.
____________
My lucky number is 75898524288+1

Profile dannyridel
Volunteer tester
Avatar
Send message
Joined: 3 Feb 19
Posts: 739
ID: 1097922
Credit: 2,595,126
RAC: 4,929
321 LLR Bronze: Earned 10,000 credits (78,095)Cullen LLR Bronze: Earned 10,000 credits (36,962)ESP LLR Bronze: Earned 10,000 credits (36,309)Generalized Cullen/Woodall LLR Bronze: Earned 10,000 credits (22,373)PPS LLR Silver: Earned 100,000 credits (177,303)PSP LLR Bronze: Earned 10,000 credits (65,677)SoB LLR Bronze: Earned 10,000 credits (66,029)SR5 LLR Bronze: Earned 10,000 credits (53,645)SGS LLR Bronze: Earned 10,000 credits (22,711)TRP LLR Bronze: Earned 10,000 credits (67,784)Woodall LLR Bronze: Earned 10,000 credits (19,031)321 Sieve Gold: Earned 500,000 credits (506,814)Generalized Cullen/Woodall Sieve (suspended) Bronze: Earned 10,000 credits (30,033)PPS Sieve Gold: Earned 500,000 credits (714,652)AP 26/27 Bronze: Earned 10,000 credits (80,860)GFN Silver: Earned 100,000 credits (243,814)PSA Silver: Earned 100,000 credits (373,034)
Message 142433 - Posted: 13 Aug 2020 | 15:30:19 UTC - in response to Message 142430.

Another problem is that any such system would think that CPUs lacking hyperthreading (typically i5) are more powerful than CPUs with hyperthreading.

Some percentage of hyperthreaded CPUs will run LLR with HT on, badly and irreversibly corrupting their performance data.

I agree with Scott, although my answer would have been shorter. Probably just one word. :)

It would be a horrendously difficult system to build and a huge effort would be required to maintain it moving forward.


I think that this kind of feature has been requested before?
____________
SHSID Electronics Group
SHSIDElectronicsGroup@outlook.com

GFN-14: 50103906^16384+1
Proth "SoB": 44243*2^440969+1


Profile NickProject donor
Avatar
Send message
Joined: 11 Jul 11
Posts: 840
ID: 105020
Credit: 1,194,793,547
RAC: 1,691,589
Discovered 3 mega primesFound 5 primes in the 2020 Tour de PrimesFound 2 mega primes in the 2020 Tour de Primes321 LLR Jade: Earned 10,000,000 credits (10,373,391)Cullen LLR Jade: Earned 10,000,000 credits (10,628,251)ESP LLR Jade: Earned 10,000,000 credits (10,142,420)Generalized Cullen/Woodall LLR Jade: Earned 10,000,000 credits (10,998,213)PPS LLR Sapphire: Earned 20,000,000 credits (20,047,078)PSP LLR Sapphire: Earned 20,000,000 credits (20,337,767)SoB LLR Sapphire: Earned 20,000,000 credits (20,827,751)SR5 LLR Jade: Earned 10,000,000 credits (10,194,757)SGS LLR Jade: Earned 10,000,000 credits (13,125,640)TRP LLR Sapphire: Earned 20,000,000 credits (20,307,419)Woodall LLR Sapphire: Earned 20,000,000 credits (22,371,625)321 Sieve Sapphire: Earned 20,000,000 credits (20,380,527)Cullen/Woodall Sieve (suspended) Gold: Earned 500,000 credits (744,531)Generalized Cullen/Woodall Sieve (suspended) Turquoise: Earned 5,000,000 credits (5,007,004)PPS Sieve Double Gold: Earned 500,000,000 credits (563,813,452)TRP Sieve (suspended) Bronze: Earned 10,000 credits (21,181)AP 26/27 Sapphire: Earned 20,000,000 credits (20,845,708)GFN Double Silver: Earned 200,000,000 credits (414,676,761)
Message 142447 - Posted: 13 Aug 2020 | 19:20:30 UTC - in response to Message 142430.

I agree with Scott, although my answer would have been shorter. Probably just one word. :)

3 words would be more entertaining. I am thinking of the quote from the Blues Brothers when Jake replies to Elwood's suggestion "We gotta go in and visit the Penguin".

Profile Rafael
Volunteer tester
Avatar
Send message
Joined: 22 Oct 14
Posts: 870
ID: 370496
Credit: 324,758,529
RAC: 125,040
321 LLR Turquoise: Earned 5,000,000 credits (7,142,710)Cullen LLR Turquoise: Earned 5,000,000 credits (7,013,155)ESP LLR Turquoise: Earned 5,000,000 credits (7,130,193)Generalized Cullen/Woodall LLR Turquoise: Earned 5,000,000 credits (7,138,464)PPS LLR Turquoise: Earned 5,000,000 credits (7,703,665)PSP LLR Turquoise: Earned 5,000,000 credits (6,995,135)SoB LLR Turquoise: Earned 5,000,000 credits (6,868,921)SR5 LLR Turquoise: Earned 5,000,000 credits (7,024,046)SGS LLR Turquoise: Earned 5,000,000 credits (7,141,730)TRP LLR Turquoise: Earned 5,000,000 credits (8,751,781)Woodall LLR Turquoise: Earned 5,000,000 credits (7,119,125)321 Sieve Jade: Earned 10,000,000 credits (10,033,828)Generalized Cullen/Woodall Sieve (suspended) Jade: Earned 10,000,000 credits (10,037,204)PPS Sieve Jade: Earned 10,000,000 credits (10,305,147)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Ruby: Earned 2,000,000 credits (2,000,053)TRP Sieve (suspended) Ruby: Earned 2,000,000 credits (2,030,160)AP 26/27 Jade: Earned 10,000,000 credits (10,253,048)GFN Sapphire: Earned 20,000,000 credits (29,312,939)PSA Double Bronze: Earned 100,000,000 credits (170,758,062)
Message 142459 - Posted: 14 Aug 2020 | 17:03:41 UTC
Last modified: 14 Aug 2020 | 17:04:40 UTC

Honestly this is the kind of stuff that can only be done with a "Primegrid Settings Calculator" à la Ryzen DRAM calculator made by the community, not with an 'official' recommendation. The way I see it, come out with a bunch of fields like Processor Gen, Core count, hyper threading and some more and spit out a couple recommendations.

Profile composite
Volunteer tester
Send message
Joined: 16 Feb 10
Posts: 764
ID: 55391
Credit: 688,170,015
RAC: 74,625
Discovered 2 mega primesFound 1 prime in the 2018 Tour de Primes321 LLR Turquoise: Earned 5,000,000 credits (5,477,467)Cullen LLR Gold: Earned 500,000 credits (776,297)ESP LLR Ruby: Earned 2,000,000 credits (3,120,351)Generalized Cullen/Woodall LLR Ruby: Earned 2,000,000 credits (2,056,207)PPS LLR Jade: Earned 10,000,000 credits (18,272,027)PSP LLR Turquoise: Earned 5,000,000 credits (5,027,818)SoB LLR Sapphire: Earned 20,000,000 credits (24,671,131)SR5 LLR Turquoise: Earned 5,000,000 credits (6,110,877)SGS LLR Ruby: Earned 2,000,000 credits (3,477,744)TRP LLR Turquoise: Earned 5,000,000 credits (7,025,303)Woodall LLR Amethyst: Earned 1,000,000 credits (1,693,614)321 Sieve Emerald: Earned 50,000,000 credits (50,256,050)Cullen/Woodall Sieve (suspended) Turquoise: Earned 5,000,000 credits (5,571,178)Generalized Cullen/Woodall Sieve (suspended) Emerald: Earned 50,000,000 credits (50,009,610)PPS Sieve Double Silver: Earned 200,000,000 credits (312,256,351)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Jade: Earned 10,000,000 credits (10,165,888)TRP Sieve (suspended) Sapphire: Earned 20,000,000 credits (20,071,454)AP 26/27 Turquoise: Earned 5,000,000 credits (6,616,128)GFN Emerald: Earned 50,000,000 credits (52,755,348)PSA Double Bronze: Earned 100,000,000 credits (102,762,384)
Message 142535 - Posted: 17 Aug 2020 | 16:11:07 UTC

This discussion devolved the wrong way.
There was no mention of GPUs in the OP.
With GPUs you pretty much get what you pay for.

I also agree that optimization is an infeasible goal.
So the OP asked for a ranking of projects, not for optimization.

I appreciate the difficulty stemming from the multitude of unknown variables.
Statistics only have significance anyway when the population size is large enough.

So the relevant question is:

Do more than 49% of hosts (not tasks), of a particular CPU type, run subprojects either so fast or so slow that the median statistic is meaningless?

Profile mikey
Avatar
Send message
Joined: 17 Mar 09
Posts: 1191
ID: 37043
Credit: 505,311,599
RAC: 131,529
Discovered 1 mega prime321 LLR Ruby: Earned 2,000,000 credits (2,038,739)Cullen LLR Ruby: Earned 2,000,000 credits (2,074,615)ESP LLR Ruby: Earned 2,000,000 credits (2,013,823)Generalized Cullen/Woodall LLR Ruby: Earned 2,000,000 credits (2,142,353)PPS LLR Turquoise: Earned 5,000,000 credits (5,149,318)PSP LLR Ruby: Earned 2,000,000 credits (2,049,284)SoB LLR Amethyst: Earned 1,000,000 credits (1,043,824)SR5 LLR Ruby: Earned 2,000,000 credits (2,053,250)SGS LLR Turquoise: Earned 5,000,000 credits (5,095,282)TRP LLR Ruby: Earned 2,000,000 credits (2,025,737)Woodall LLR Amethyst: Earned 1,000,000 credits (1,893,227)321 Sieve Sapphire: Earned 20,000,000 credits (23,770,672)Cullen/Woodall Sieve (suspended) Gold: Earned 500,000 credits (944,431)Generalized Cullen/Woodall Sieve (suspended) Sapphire: Earned 20,000,000 credits (20,813,253)PPS Sieve Double Silver: Earned 200,000,000 credits (339,648,557)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Ruby: Earned 2,000,000 credits (2,446,797)AP 26/27 Sapphire: Earned 20,000,000 credits (33,140,471)GFN Sapphire: Earned 20,000,000 credits (36,517,558)PSA Sapphire: Earned 20,000,000 credits (20,457,430)
Message 142574 - Posted: 18 Aug 2020 | 3:11:08 UTC - in response to Message 142535.

This discussion devolved the wrong way.
There was no mention of GPUs in the OP.
With GPUs you pretty much get what you pay for.

I also agree that optimization is an infeasible goal.
So the OP asked for a ranking of projects, not for optimization.

I appreciate the difficulty stemming from the multitude of unknown variables.
Statistics only have significance anyway when the population size is large enough.

So the relevant question is:

Do more than 49% of hosts (not tasks), of a particular CPU type, run subprojects either so fast or so slow that the median statistic is meaningless?


Very well said!! And if the answer is NO can the Project come up with a way to state that in a way the average user can use it to crunch better than the user who's just winging it? I don't really think anyone thinks most users can optimize their machines to the nth degree, most won't even try, but can we be better than we are and does the Project have the info we can use to do that is the question? Will the data still be 'generic' yes it will but if for example I have an Intel 3930 cpu and I'm running 6 tasks at one time can I do better at the LLR tasks? How about the Sieve tasks? I think that data is in their someplace the question is can it be gotten without 100 hours of data mining that can best be used elsewhere?

Scott BrownProject donor
Volunteer moderator
Project administrator
Volunteer tester
Project scientist
Avatar
Send message
Joined: 17 Oct 05
Posts: 2125
ID: 1178
Credit: 8,410,720,955
RAC: 6,716,484
Discovered the World's First base 116 Generalized Cullen prime!!!Discovered 26 mega primesEliminated 7 conjecture "k"sDiscovered 1 Sophie Germain pairDiscovered 2 Fermat divisors2012 Tour de Primes highest prime count2012 Tour de Primes most Mountain Stage primes2015 Tour de Primes highest prime count2016 Tour de Primes highest prime countFound 23 primes in the 2018 Tour de PrimesFound 1 mega prime in the 2018 Tour de PrimesFound 2 primes in the 2018 Tour de Primes Mountain Stage2019 Tour de Primes highest prime countFound 22 primes in the 2019 Tour de Primes2020 Tour de Primes highest prime scoreFound 21 primes in the 2020 Tour de PrimesFound 4 mega primes in the 2020 Tour de Primes321 LLR Double Bronze: Earned 100,000,000 credits (185,327,043)Cullen LLR Double Bronze: Earned 100,000,000 credits (103,870,990)ESP LLR Double Silver: Earned 200,000,000 credits (203,249,784)Generalized Cullen/Woodall LLR Double Bronze: Earned 100,000,000 credits (109,580,172)PPS LLR Double Gold: Earned 500,000,000 credits (639,557,324)PSP LLR Double Bronze: Earned 100,000,000 credits (126,982,721)SoB LLR Double Bronze: Earned 100,000,000 credits (135,747,083)SR5 LLR Double Silver: Earned 200,000,000 credits (214,194,272)SGS LLR Double Silver: Earned 200,000,000 credits (200,486,586)TPS LLR (retired) Silver: Earned 100,000 credits (235,439)TRP LLR Double Silver: Earned 200,000,000 credits (201,215,056)Woodall LLR Double Bronze: Earned 100,000,000 credits (101,447,725)321 Sieve Double Silver: Earned 200,000,000 credits (235,451,253)Cullen/Woodall Sieve (suspended) Emerald: Earned 50,000,000 credits (83,794,448)Generalized Cullen/Woodall Sieve (suspended) Double Silver: Earned 200,000,000 credits (285,139,652)PPS Sieve Double Ruby: Earned 2,000,000,000 credits (2,665,703,405)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Double Silver: Earned 200,000,000 credits (203,523,358)TRP Sieve (suspended) Double Silver: Earned 200,000,000 credits (201,489,157)AP 26/27 Double Silver: Earned 200,000,000 credits (374,978,656)GFN Double Amethyst: Earned 1,000,000,000 credits (1,879,778,210)PSA Double Silver: Earned 200,000,000 credits (259,058,048)
Message 142577 - Posted: 18 Aug 2020 | 12:27:41 UTC - in response to Message 142574.
Last modified: 18 Aug 2020 | 12:33:25 UTC

This discussion devolved the wrong way.
There was no mention of GPUs in the OP.
With GPUs you pretty much get what you pay for.

I also agree that optimization is an infeasible goal.
So the OP asked for a ranking of projects, not for optimization.

I appreciate the difficulty stemming from the multitude of unknown variables.
Statistics only have significance anyway when the population size is large enough.

So the relevant question is:

Do more than 49% of hosts (not tasks), of a particular CPU type, run subprojects either so fast or so slow that the median statistic is meaningless?




Very well said!! And if the answer is NO can the Project come up with a way to state that in a way the average user can use it to crunch better than the user who's just winging it? I don't really think anyone thinks most users can optimize their machines to the nth degree, most won't even try, but can we be better than we are and does the Project have the info we can use to do that is the question? Will the data still be 'generic' yes it will but if for example I have an Intel 3930 cpu and I'm running 6 tasks at one time can I do better at the LLR tasks? How about the Sieve tasks? I think that data is in their someplace the question is can it be gotten without 100 hours of data mining that can best be used elsewhere?


I apologize if I sound frustrated in this response, but I am getting to that point a bit. I did not use the GPU example in response to the OP's specific call, but rather as an example (which I thought was easier to demonstrate than for CPU) regarding why BOINC statistics/machine info are largely useless for what is being asked.

So let's try this one more time...Can the project produce a meaningful subproject ranking to show which ones are best for particular CPUs? NO. That is because 1) much of the information to do so accurately is not collected and 2) what is collected is VERY unreliable due to the variety of conditions that apply to the data for each specific CPU. For example, lets suppose that we have a few i7-3930 CPUs to gather timing data from across all the projects. One of those might be running 6-core mutlithread tasks on all projects it does with HT on and 50% CPU settings, another might be doing the same mulithread setting with HT off, another might be running 12-threaded, another might be running 2x 3-threaded, another is running single-threaded with HT off, etc. Throw in the fact that some of those have no GPU, some have a GPU not running BOINC, some have a GPU doing BOINC, some have multi-GPUs with some mixture of work, etc. Unlike the GPU reporting, BOINC does not record how many CPUs are in a system, so for some CPUs the reporting of times may reflect computational efficiencies of 2 or more CPUs doing work on a single unit. Now add in that some of those CPUs are running more than one subproject at a time with a mixture of work, and those that are are not all running the same mix of work. And then we have the fact that some of the subprojects have different FFTs at the same time, which can vary substantially (e.g., TRP has FFT values ranging from about 768k to 1152k right now) and produce drastically different performance numbers on the same CPU. PG work times are also directly affected by the other work being done on a CPU, and BOINC has no way to track this (and while many might think of it as random noise, it is not--e.g., higher end systems would be more likely to be affected than older, slower systems since the former would be more likely to be running high intensity applications such as games, design software, etc.). These are but a few examples of the variables invovled in producing an accurate ranking, many if not most we have no access to, and even if we did it would still be quite complex to figure out an accurate ranking.

So basically, if you want to know which projects are best for your CPU, you will have FAR BETTER data running a few tests yourself than we could ever produce with the data that we get from the BOINC system.

In general, you can also do pretty well with just some basic guidelines...

1) GPU projects (GFN, AP, PPS Sieve) are best on GPU, period. CPUs are always suboptimal on these.
2) LLR tasks are best with modern CPUs--AVX512 is best (if you have two AVX512 units), FMA3/AVX2 is next best, and that is followed by AVX.
3) Pre-AVX CPUs tend to do best on the CPU sieves.
4) FFT size matters...do work that fits inside the CPU cache as much as possible for best performance. A rough estimate is to do FFTx8 for the MB size in cache (e.g., an FFT of 1M will use about 8MB of cache, etc.).

Profile j.sheridanProject donor
Send message
Joined: 21 Mar 11
Posts: 659
ID: 91622
Credit: 1,108,930,436
RAC: 2,168,048
321 LLR Turquoise: Earned 5,000,000 credits (5,019,701)Cullen LLR Ruby: Earned 2,000,000 credits (4,582,271)ESP LLR Ruby: Earned 2,000,000 credits (4,563,800)Generalized Cullen/Woodall LLR Ruby: Earned 2,000,000 credits (3,975,172)PPS LLR Turquoise: Earned 5,000,000 credits (8,639,452)PSP LLR Turquoise: Earned 5,000,000 credits (5,500,951)SoB LLR Turquoise: Earned 5,000,000 credits (8,583,226)SR5 LLR Ruby: Earned 2,000,000 credits (3,441,728)SGS LLR Turquoise: Earned 5,000,000 credits (5,169,579)TRP LLR Turquoise: Earned 5,000,000 credits (9,717,538)Woodall LLR Ruby: Earned 2,000,000 credits (4,084,622)321 Sieve Sapphire: Earned 20,000,000 credits (20,019,388)Cullen/Woodall Sieve (suspended) Double Silver: Earned 200,000,000 credits (265,102,350)PPS Sieve Double Gold: Earned 500,000,000 credits (514,289,554)TRP Sieve (suspended) Turquoise: Earned 5,000,000 credits (5,801,812)AP 26/27 Emerald: Earned 50,000,000 credits (53,472,718)GFN Double Bronze: Earned 100,000,000 credits (186,978,187)
Message 142587 - Posted: 18 Aug 2020 | 16:01:49 UTC - in response to Message 142577.

What might be helpful would be to have a sticky on how to use prime95 benchmarking to find the best option for your cpu for a given FFT size and then list the current FFT sizes for each sub-project on the selection screen with a "* Please check the thread if you want to get the most efficient use out of your cpu".

Profile dannyridel
Volunteer tester
Avatar
Send message
Joined: 3 Feb 19
Posts: 739
ID: 1097922
Credit: 2,595,126
RAC: 4,929
321 LLR Bronze: Earned 10,000 credits (78,095)Cullen LLR Bronze: Earned 10,000 credits (36,962)ESP LLR Bronze: Earned 10,000 credits (36,309)Generalized Cullen/Woodall LLR Bronze: Earned 10,000 credits (22,373)PPS LLR Silver: Earned 100,000 credits (177,303)PSP LLR Bronze: Earned 10,000 credits (65,677)SoB LLR Bronze: Earned 10,000 credits (66,029)SR5 LLR Bronze: Earned 10,000 credits (53,645)SGS LLR Bronze: Earned 10,000 credits (22,711)TRP LLR Bronze: Earned 10,000 credits (67,784)Woodall LLR Bronze: Earned 10,000 credits (19,031)321 Sieve Gold: Earned 500,000 credits (506,814)Generalized Cullen/Woodall Sieve (suspended) Bronze: Earned 10,000 credits (30,033)PPS Sieve Gold: Earned 500,000 credits (714,652)AP 26/27 Bronze: Earned 10,000 credits (80,860)GFN Silver: Earned 100,000 credits (243,814)PSA Silver: Earned 100,000 credits (373,034)
Message 142592 - Posted: 19 Aug 2020 | 2:42:23 UTC - in response to Message 142587.

What might be helpful would be to have a sticky on how to use prime95 benchmarking to find the best option for your cpu for a given FFT size and then list the current FFT sizes for each sub-project on the selection screen with a "* Please check the thread if you want to get the most efficient use out of your cpu".

here
____________
SHSID Electronics Group
SHSIDElectronicsGroup@outlook.com

GFN-14: 50103906^16384+1
Proth "SoB": 44243*2^440969+1


Profile j.sheridanProject donor
Send message
Joined: 21 Mar 11
Posts: 659
ID: 91622
Credit: 1,108,930,436
RAC: 2,168,048
321 LLR Turquoise: Earned 5,000,000 credits (5,019,701)Cullen LLR Ruby: Earned 2,000,000 credits (4,582,271)ESP LLR Ruby: Earned 2,000,000 credits (4,563,800)Generalized Cullen/Woodall LLR Ruby: Earned 2,000,000 credits (3,975,172)PPS LLR Turquoise: Earned 5,000,000 credits (8,639,452)PSP LLR Turquoise: Earned 5,000,000 credits (5,500,951)SoB LLR Turquoise: Earned 5,000,000 credits (8,583,226)SR5 LLR Ruby: Earned 2,000,000 credits (3,441,728)SGS LLR Turquoise: Earned 5,000,000 credits (5,169,579)TRP LLR Turquoise: Earned 5,000,000 credits (9,717,538)Woodall LLR Ruby: Earned 2,000,000 credits (4,084,622)321 Sieve Sapphire: Earned 20,000,000 credits (20,019,388)Cullen/Woodall Sieve (suspended) Double Silver: Earned 200,000,000 credits (265,102,350)PPS Sieve Double Gold: Earned 500,000,000 credits (514,289,554)TRP Sieve (suspended) Turquoise: Earned 5,000,000 credits (5,801,812)AP 26/27 Emerald: Earned 50,000,000 credits (53,472,718)GFN Double Bronze: Earned 100,000,000 credits (186,978,187)
Message 142593 - Posted: 19 Aug 2020 | 7:44:07 UTC - in response to Message 142592.

What might be helpful would be to have a sticky on how to use prime95 benchmarking to find the best option for your cpu for a given FFT size and then list the current FFT sizes for each sub-project on the selection screen with a "* Please check the thread if you want to get the most efficient use out of your cpu".

here

ok we have a thread which is good but the most important thing is to make it obvious where to find it.
I didn't manage to find anything when I went searching a few months ago so did my testing by running units with different settings.
It was only by luck a while after that that I found a thread where someone mentioned processor affinity for Ryzen. This is critical to getting the best out of these CPUs when running multi-threaded tasks.

Profile dannyridel
Volunteer tester
Avatar
Send message
Joined: 3 Feb 19
Posts: 739
ID: 1097922
Credit: 2,595,126
RAC: 4,929
321 LLR Bronze: Earned 10,000 credits (78,095)Cullen LLR Bronze: Earned 10,000 credits (36,962)ESP LLR Bronze: Earned 10,000 credits (36,309)Generalized Cullen/Woodall LLR Bronze: Earned 10,000 credits (22,373)PPS LLR Silver: Earned 100,000 credits (177,303)PSP LLR Bronze: Earned 10,000 credits (65,677)SoB LLR Bronze: Earned 10,000 credits (66,029)SR5 LLR Bronze: Earned 10,000 credits (53,645)SGS LLR Bronze: Earned 10,000 credits (22,711)TRP LLR Bronze: Earned 10,000 credits (67,784)Woodall LLR Bronze: Earned 10,000 credits (19,031)321 Sieve Gold: Earned 500,000 credits (506,814)Generalized Cullen/Woodall Sieve (suspended) Bronze: Earned 10,000 credits (30,033)PPS Sieve Gold: Earned 500,000 credits (714,652)AP 26/27 Bronze: Earned 10,000 credits (80,860)GFN Silver: Earned 100,000 credits (243,814)PSA Silver: Earned 100,000 credits (373,034)
Message 142594 - Posted: 19 Aug 2020 | 8:29:31 UTC - in response to Message 142593.

What might be helpful would be to have a sticky on how to use prime95 benchmarking to find the best option for your cpu for a given FFT size and then list the current FFT sizes for each sub-project on the selection screen with a "* Please check the thread if you want to get the most efficient use out of your cpu".

here

ok we have a thread which is good but the most important thing is to make it obvious where to find it.
I didn't manage to find anything when I went searching a few months ago so did my testing by running units with different settings.
It was only by luck a while after that that I found a thread where someone mentioned processor affinity for Ryzen. This is critical to getting the best out of these CPUs when running multi-threaded tasks.


I subscribe to these useful threads. The affinity thread was started when I was banned, but I still could subscribe to it. Its useful. ;)

____________
SHSID Electronics Group
SHSIDElectronicsGroup@outlook.com

GFN-14: 50103906^16384+1
Proth "SoB": 44243*2^440969+1


Profile firedrakes
Avatar
Send message
Joined: 7 Feb 09
Posts: 55
ID: 35268
Credit: 56,864,740
RAC: 29,031
321 LLR Gold: Earned 500,000 credits (878,090)Cullen LLR Bronze: Earned 10,000 credits (56,880)ESP LLR Gold: Earned 500,000 credits (920,612)Generalized Cullen/Woodall LLR Silver: Earned 100,000 credits (273,624)PPS LLR Turquoise: Earned 5,000,000 credits (5,038,960)PSP LLR Silver: Earned 100,000 credits (104,325)SoB LLR Silver: Earned 100,000 credits (109,708)SR5 LLR Gold: Earned 500,000 credits (983,230)SGS LLR Ruby: Earned 2,000,000 credits (3,117,946)TRP LLR Silver: Earned 100,000 credits (253,030)Woodall LLR Silver: Earned 100,000 credits (117,458)321 Sieve Turquoise: Earned 5,000,000 credits (7,602,200)Cullen/Woodall Sieve (suspended) Silver: Earned 100,000 credits (480,693)Generalized Cullen/Woodall Sieve (suspended) Amethyst: Earned 1,000,000 credits (1,644,320)PPS Sieve Turquoise: Earned 5,000,000 credits (7,270,014)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Bronze: Earned 10,000 credits (16,303)AP 26/27 Amethyst: Earned 1,000,000 credits (1,537,276)GFN Sapphire: Earned 20,000,000 credits (26,454,089)
Message 142609 - Posted: 20 Aug 2020 | 9:52:55 UTC - in response to Message 142417.

i love this to. it would really help out

Profile Pooh Bear 27Project donor
Avatar
Send message
Joined: 10 May 09
Posts: 697
ID: 39821
Credit: 406,835,310
RAC: 20,743
Discovered 1 mega primeEliminated 1 conjecture "k"Discovered 1 twin prime321 LLR Turquoise: Earned 5,000,000 credits (5,014,405)Cullen LLR Turquoise: Earned 5,000,000 credits (5,734,854)ESP LLR Turquoise: Earned 5,000,000 credits (5,421,149)Generalized Cullen/Woodall LLR Turquoise: Earned 5,000,000 credits (5,576,209)PPS LLR Jade: Earned 10,000,000 credits (17,389,837)PSP LLR Jade: Earned 10,000,000 credits (10,425,342)SoB LLR Turquoise: Earned 5,000,000 credits (8,986,628)SR5 LLR Turquoise: Earned 5,000,000 credits (5,248,659)SGS LLR Turquoise: Earned 5,000,000 credits (6,878,856)TRP LLR Turquoise: Earned 5,000,000 credits (5,416,106)Woodall LLR Turquoise: Earned 5,000,000 credits (6,400,237)321 Sieve Amethyst: Earned 1,000,000 credits (1,000,576)Cullen/Woodall Sieve (suspended) Ruby: Earned 2,000,000 credits (4,000,977)Generalized Cullen/Woodall Sieve (suspended) Turquoise: Earned 5,000,000 credits (6,843,367)PPS Sieve Emerald: Earned 50,000,000 credits (78,673,438)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Turquoise: Earned 5,000,000 credits (6,473,750)TRP Sieve (suspended) Turquoise: Earned 5,000,000 credits (8,243,608)AP 26/27 Jade: Earned 10,000,000 credits (11,872,374)GFN Double Silver: Earned 200,000,000 credits (202,192,765)PSA Turquoise: Earned 5,000,000 credits (5,042,125)
Message 142613 - Posted: 20 Aug 2020 | 13:46:13 UTC - in response to Message 142609.

i love this to. it would really help out

Did you read the thread? Not something that can be done easily. Please read the whole thread and learn.

Profile firedrakes
Avatar
Send message
Joined: 7 Feb 09
Posts: 55
ID: 35268
Credit: 56,864,740
RAC: 29,031
321 LLR Gold: Earned 500,000 credits (878,090)Cullen LLR Bronze: Earned 10,000 credits (56,880)ESP LLR Gold: Earned 500,000 credits (920,612)Generalized Cullen/Woodall LLR Silver: Earned 100,000 credits (273,624)PPS LLR Turquoise: Earned 5,000,000 credits (5,038,960)PSP LLR Silver: Earned 100,000 credits (104,325)SoB LLR Silver: Earned 100,000 credits (109,708)SR5 LLR Gold: Earned 500,000 credits (983,230)SGS LLR Ruby: Earned 2,000,000 credits (3,117,946)TRP LLR Silver: Earned 100,000 credits (253,030)Woodall LLR Silver: Earned 100,000 credits (117,458)321 Sieve Turquoise: Earned 5,000,000 credits (7,602,200)Cullen/Woodall Sieve (suspended) Silver: Earned 100,000 credits (480,693)Generalized Cullen/Woodall Sieve (suspended) Amethyst: Earned 1,000,000 credits (1,644,320)PPS Sieve Turquoise: Earned 5,000,000 credits (7,270,014)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Bronze: Earned 10,000 credits (16,303)AP 26/27 Amethyst: Earned 1,000,000 credits (1,537,276)GFN Sapphire: Earned 20,000,000 credits (26,454,089)
Message 142684 - Posted: 25 Aug 2020 | 2:08:35 UTC - in response to Message 142613.

i did.
can it be done. yes ..

but wont due to time or money.

this boinc project is the hardest one of them all. so their is a novice entrance bearer .
even a soft case would be helpful.

Tern
Volunteer developer
Volunteer tester
Avatar
Send message
Joined: 20 Sep 15
Posts: 26
ID: 421148
Credit: 292,405,082
RAC: 531,822
Found 1 prime in the 2020 Tour de Primes321 LLR Ruby: Earned 2,000,000 credits (2,133,351)Cullen LLR Turquoise: Earned 5,000,000 credits (5,420,696)ESP LLR Turquoise: Earned 5,000,000 credits (9,319,231)Generalized Cullen/Woodall LLR Ruby: Earned 2,000,000 credits (2,954,535)PPS LLR Jade: Earned 10,000,000 credits (18,734,385)PSP LLR Jade: Earned 10,000,000 credits (10,085,457)SoB LLR Turquoise: Earned 5,000,000 credits (5,346,126)SR5 LLR Ruby: Earned 2,000,000 credits (2,280,539)SGS LLR Amethyst: Earned 1,000,000 credits (1,028,652)TRP LLR Turquoise: Earned 5,000,000 credits (6,980,710)Woodall LLR Turquoise: Earned 5,000,000 credits (5,243,846)321 Sieve Sapphire: Earned 20,000,000 credits (27,356,392)Generalized Cullen/Woodall Sieve (suspended) Jade: Earned 10,000,000 credits (11,609,366)PPS Sieve Emerald: Earned 50,000,000 credits (99,437,758)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Silver: Earned 100,000 credits (128,472)TRP Sieve (suspended) Gold: Earned 500,000 credits (631,827)AP 26/27 Sapphire: Earned 20,000,000 credits (21,448,115)GFN Emerald: Earned 50,000,000 credits (62,254,185)PSA Bronze: Earned 10,000 credits (15,807)
Message 143263 - Posted: 13 Sep 2020 | 6:42:12 UTC

As a Systems Analyst for many years, I always hate to hear (or say) "We can't do that." But in this case, while there are lots of fields in the database that LOOK like they'd be useful, BOINC doesn't collect the data in a consistent, accurate manner. If we did have the time and money to do this analysis, the results would (99.5% likely) be useless. Garbage in - garbage out.

That said, the goal COULD be reached (with enough time and money) - I'd simply write a standalone program that the user could download and run on THEIR SPECIFIC machine, that would run just enough PG-like tasks within it to get the good numbers you'd want, and then spew out recommendations. I can find the programmers with the time, if someone wants to come up with the money! :-)

Honestly, while I'd love to see this myself, I don't think it's worth the investment it would take.

urbanknight
Send message
Joined: 10 Jan 18
Posts: 3
ID: 967824
Credit: 14,265,264
RAC: 21,881
321 LLR Amethyst: Earned 1,000,000 credits (1,145,373)Cullen LLR Amethyst: Earned 1,000,000 credits (1,024,165)ESP LLR Amethyst: Earned 1,000,000 credits (1,184,836)Generalized Cullen/Woodall LLR Amethyst: Earned 1,000,000 credits (1,395,630)PPS LLR Amethyst: Earned 1,000,000 credits (1,407,307)PSP LLR Gold: Earned 500,000 credits (678,341)SoB LLR Gold: Earned 500,000 credits (760,528)SR5 LLR Gold: Earned 500,000 credits (514,798)SGS LLR Amethyst: Earned 1,000,000 credits (1,362,090)TRP LLR Gold: Earned 500,000 credits (763,259)Woodall LLR Amethyst: Earned 1,000,000 credits (1,043,315)321 Sieve Gold: Earned 500,000 credits (611,715)Generalized Cullen/Woodall Sieve (suspended) Gold: Earned 500,000 credits (542,240)PPS Sieve Gold: Earned 500,000 credits (718,023)AP 26/27 Gold: Earned 500,000 credits (537,719)GFN Gold: Earned 500,000 credits (575,924)
Message 143339 - Posted: 16 Sep 2020 | 9:05:18 UTC

My opinion (if someone needs it) is to draw conclusions about the most effective setting based on some input computer parameters - a failure idea. Too many fields and too many nuances. Take a look at the processors, even among the same model there are ten to twenty variants with different configurations, this is not talking about the options when an application that always hangs in the tray can affect performance.
The correct approach here is to evaluate the performance of this particular machine, with this particular configuration hardware + software + tasks in addition to BOINC.
The correct solution is an additional assessment program, inside which there are examples of tasks that need to be solved. The program launches a task with some settings, the same task with others, with the third, and so on, until it goes over all or the most interesting ones, and then, based on the statistics of solving tasks by this particular computer, it concludes that on this computer, from here It is better to solve such a configuration here this project, load 2 tasks, 4 cores per task.

Post to thread

Message boards : Number crunching : Automatic subproject recommendation for each host

[Return to PrimeGrid main page]
DNS Powered by DNSEXIT.COM
Copyright © 2005 - 2020 Rytis Slatkevičius (contact) and PrimeGrid community. Server load 2.94, 2.22, 2.00
Generated 31 Oct 2020 | 13:54:35 UTC