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 : Proth Prime Search : Why PPS,Mega PPS, PPSE using FMA3 but non AVX2?

Author Message
RobertKazan
Send message
Joined: 8 Feb 12
Posts: 23
ID: 129951
Credit: 19,928,297
RAC: 94,617
ESP LLR Bronze: Earned 10,000 credits (55,289)PPS LLR Ruby: Earned 2,000,000 credits (4,985,748)SoB LLR Silver: Earned 100,000 credits (378,617)SR5 LLR Silver: Earned 100,000 credits (244,237)SGS LLR Gold: Earned 500,000 credits (819,053)TRP LLR Bronze: Earned 10,000 credits (14,750)321 Sieve Gold: Earned 500,000 credits (601,956)Cullen/Woodall Sieve (suspended) Gold: Earned 500,000 credits (815,330)PPS Sieve Turquoise: Earned 5,000,000 credits (5,447,536)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Silver: Earned 100,000 credits (371,711)TRP Sieve (suspended) Bronze: Earned 10,000 credits (64,330)GFN Turquoise: Earned 5,000,000 credits (6,118,889)
Message 99069 - Posted: 23 Sep 2016 | 4:55:59 UTC
Last modified: 23 Sep 2016 | 5:12:53 UTC

Why PPS,Mega PPS, PPSE using FMA3 but non AVX2 on Intel Haswell and later processors? unless these instructions are not faster than FMA3?

mackerelProject donor
Volunteer tester
Avatar
Send message
Joined: 2 Oct 08
Posts: 2234
ID: 29980
Credit: 340,818,302
RAC: 349,851
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 Primes321 LLR Turquoise: Earned 5,000,000 credits (8,743,303)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 Sapphire: Earned 20,000,000 credits (49,080,462)PSP LLR Jade: Earned 10,000,000 credits (14,111,955)SoB LLR Jade: Earned 10,000,000 credits (17,319,914)SR5 LLR Jade: Earned 10,000,000 credits (18,479,395)SGS LLR Turquoise: Earned 5,000,000 credits (6,702,766)TPS LLR (retired) Bronze: Earned 10,000 credits (34,130)TRP LLR Jade: Earned 10,000,000 credits (15,154,737)Woodall LLR Turquoise: Earned 5,000,000 credits (5,715,464)321 Sieve Turquoise: Earned 5,000,000 credits (8,326,385)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 (74,140,875)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 Sapphire: Earned 20,000,000 credits (48,107,585)PSA Ruby: Earned 2,000,000 credits (2,939,755)
Message 99072 - Posted: 23 Sep 2016 | 6:22:05 UTC - in response to Message 99069.

AVX2 are integer instructions. Currently the heavy work is done by floating point instructions, in the form of FMA3 where appropriate. I don't understand the maths sufficiently to say why that is preferable in performance terms. If anyone can come up with something faster than current, whatever the instructions used, there will be a lot of people interested.

RobertKazan
Send message
Joined: 8 Feb 12
Posts: 23
ID: 129951
Credit: 19,928,297
RAC: 94,617
ESP LLR Bronze: Earned 10,000 credits (55,289)PPS LLR Ruby: Earned 2,000,000 credits (4,985,748)SoB LLR Silver: Earned 100,000 credits (378,617)SR5 LLR Silver: Earned 100,000 credits (244,237)SGS LLR Gold: Earned 500,000 credits (819,053)TRP LLR Bronze: Earned 10,000 credits (14,750)321 Sieve Gold: Earned 500,000 credits (601,956)Cullen/Woodall Sieve (suspended) Gold: Earned 500,000 credits (815,330)PPS Sieve Turquoise: Earned 5,000,000 credits (5,447,536)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Silver: Earned 100,000 credits (371,711)TRP Sieve (suspended) Bronze: Earned 10,000 credits (64,330)GFN Turquoise: Earned 5,000,000 credits (6,118,889)
Message 99073 - Posted: 23 Sep 2016 | 6:26:45 UTC
Last modified: 23 Sep 2016 | 6:32:39 UTC

Thank you. But LinX 0.6.5 AVX2 shows approximately 2 times leap than AVX.
My core i5 2500k@4000Mhz shows same time as Core i5 6600k.

mackerelProject donor
Volunteer tester
Avatar
Send message
Joined: 2 Oct 08
Posts: 2234
ID: 29980
Credit: 340,818,302
RAC: 349,851
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 Primes321 LLR Turquoise: Earned 5,000,000 credits (8,743,303)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 Sapphire: Earned 20,000,000 credits (49,080,462)PSP LLR Jade: Earned 10,000,000 credits (14,111,955)SoB LLR Jade: Earned 10,000,000 credits (17,319,914)SR5 LLR Jade: Earned 10,000,000 credits (18,479,395)SGS LLR Turquoise: Earned 5,000,000 credits (6,702,766)TPS LLR (retired) Bronze: Earned 10,000 credits (34,130)TRP LLR Jade: Earned 10,000,000 credits (15,154,737)Woodall LLR Turquoise: Earned 5,000,000 credits (5,715,464)321 Sieve Turquoise: Earned 5,000,000 credits (8,326,385)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 (74,140,875)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 Sapphire: Earned 20,000,000 credits (48,107,585)PSA Ruby: Earned 2,000,000 credits (2,939,755)
Message 99075 - Posted: 23 Sep 2016 | 6:45:52 UTC - in response to Message 99073.

I can't comment on what LinX does, and if it is applicable here. There may be some confusion also, in that although the instructions used are FMA, sometimes they get lumped in with AVX2 since they were introduced at the same time.

I don't see a 6600k listed in your computer list. Have you ran actual units through it? You should see a nice speed out of the 6600k, depending on what clock it is running at.

If we assume for now that the 6600k was overclocked to 4 GHz like the 2500k, I would expect it to see about 1.7x faster runtime on it for PPS, PPSE and SGS units. Not MEGA as that is getting bigger and will start to be limited by ram speed.

In case you wonder why, Sandy Bridge (2500k) didn't support FMA which was introduced with Haswell. In my testing this is about 50% per clock faster so it gives a big boost. Skylake (6600k) didn't add any useful new instructions, but they did something to make it more efficient overall and it is about 14% faster at Prime tasks than Haswell. Combine the two for about 71% speed increase (the boosts are multiplied, not added).

axn
Volunteer developer
Send message
Joined: 29 Dec 07
Posts: 285
ID: 16874
Credit: 28,027,106
RAC: 0
321 LLR Gold: Earned 500,000 credits (656,177)Cullen LLR Silver: Earned 100,000 credits (457,297)ESP LLR Silver: Earned 100,000 credits (437,502)PPS LLR Amethyst: Earned 1,000,000 credits (1,256,334)SoB LLR Amethyst: Earned 1,000,000 credits (1,246,109)SR5 LLR Ruby: Earned 2,000,000 credits (2,193,129)SGS LLR Silver: Earned 100,000 credits (101,658)Woodall LLR Silver: Earned 100,000 credits (292,129)Cullen/Woodall Sieve (suspended) Bronze: Earned 10,000 credits (32,498)PPS Sieve Jade: Earned 10,000,000 credits (15,515,229)AP 26/27 Amethyst: Earned 1,000,000 credits (1,491,867)GFN Ruby: Earned 2,000,000 credits (4,245,870)PSA Silver: Earned 100,000 credits (101,309)
Message 99076 - Posted: 23 Sep 2016 | 7:48:37 UTC

https://en.wikipedia.org/wiki/Advanced_Vector_Extensions#Advanced_Vector_Extensions_2

FMA3 is part of AVX2!

stream
Volunteer developer
Volunteer tester
Send message
Joined: 1 Mar 14
Posts: 572
ID: 301928
Credit: 451,695,603
RAC: 1,769
Discovered 1 mega primeFound 1 prime in the 2018 Tour de PrimesFound 1 prime in the 2019 Tour de Primes321 LLR Turquoise: Earned 5,000,000 credits (9,919,609)Cullen LLR Turquoise: Earned 5,000,000 credits (9,934,320)ESP LLR Turquoise: Earned 5,000,000 credits (9,909,084)Generalized Cullen/Woodall LLR Turquoise: Earned 5,000,000 credits (5,921,052)PPS LLR Turquoise: Earned 5,000,000 credits (7,262,900)PSP LLR Turquoise: Earned 5,000,000 credits (5,089,560)SoB LLR Turquoise: Earned 5,000,000 credits (5,824,522)SR5 LLR Turquoise: Earned 5,000,000 credits (5,399,087)SGS LLR Turquoise: Earned 5,000,000 credits (5,435,504)TRP LLR Turquoise: Earned 5,000,000 credits (9,911,706)Woodall LLR Turquoise: Earned 5,000,000 credits (5,011,851)321 Sieve Sapphire: Earned 20,000,000 credits (20,004,228)Generalized Cullen/Woodall Sieve (suspended) Sapphire: Earned 20,000,000 credits (20,047,667)PPS Sieve Sapphire: Earned 20,000,000 credits (20,866,490)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Sapphire: Earned 20,000,000 credits (20,043,271)TRP Sieve (suspended) Sapphire: Earned 20,000,000 credits (20,015,177)AP 26/27 Sapphire: Earned 20,000,000 credits (20,045,194)GFN Emerald: Earned 50,000,000 credits (50,752,940)PSA Double Silver: Earned 200,000,000 credits (200,301,443)
Message 99077 - Posted: 23 Sep 2016 | 9:18:26 UTC - in response to Message 99069.

Why PPS,Mega PPS, PPSE using FMA3 but non AVX2 on Intel Haswell and later processors? unless these instructions are not faster than FMA3?

FMA it's not a instruction set, is just a single instruction which calculates r=a*b+c in single fast operation, which is very useful for some types of mathematical calculations. FMA is a extension of AVX2 instruction set. Although FMA and AVX2 could be enabled separately in CPU, a program which uses FMA instruction usually assumes that AVX2 instructions are also available and use them, if it's beneficial for program.

RobertKazan
Send message
Joined: 8 Feb 12
Posts: 23
ID: 129951
Credit: 19,928,297
RAC: 94,617
ESP LLR Bronze: Earned 10,000 credits (55,289)PPS LLR Ruby: Earned 2,000,000 credits (4,985,748)SoB LLR Silver: Earned 100,000 credits (378,617)SR5 LLR Silver: Earned 100,000 credits (244,237)SGS LLR Gold: Earned 500,000 credits (819,053)TRP LLR Bronze: Earned 10,000 credits (14,750)321 Sieve Gold: Earned 500,000 credits (601,956)Cullen/Woodall Sieve (suspended) Gold: Earned 500,000 credits (815,330)PPS Sieve Turquoise: Earned 5,000,000 credits (5,447,536)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Silver: Earned 100,000 credits (371,711)TRP Sieve (suspended) Bronze: Earned 10,000 credits (64,330)GFN Turquoise: Earned 5,000,000 credits (6,118,889)
Message 99084 - Posted: 23 Sep 2016 | 20:37:02 UTC

Thanks for your questions.

Message boards : Proth Prime Search : Why PPS,Mega PPS, PPSE using FMA3 but non AVX2?

[Return to PrimeGrid main page]
DNS Powered by DNSEXIT.COM
Copyright © 2005 - 2019 Rytis Slatkevičius (contact) and PrimeGrid community. Server load 2.58, 1.97, 1.88
Generated 12 Nov 2019 | 8:25:27 UTC