What happens with my hyperthreading?

[syzygy]

Hmm, so what's left? Windows SMT and maybe some kind of CPU micro-code update?

--
Srdja

Some nonsense is left. I will restart my PC after I finish today my project.

Do you have any old nps numbers? It seems quite possible that all your nps numbers are lower now due to microcode updates to fix security issues. The hyperthreading ones are just less lower.

[MikeB]

With mine i7 4790 4 physical cores, 8 logical, I seem to get a tremendous boost from hyperthreading with both SF_dev and SF NNUE, especially SF NNUE. Did something happened with Intel hyperthreading on Windows 10 and I am unaware of?

First, SF_dev in ultra-fast games 8 threads vs 4 threads with LittleBlitzer, to measure NPS and similar things:

Code: Select all

Games Completed = 1000 of 1000 (Avg game length = 6.162 sec)
Settings = Gauntlet/128MB/1500ms+25ms/M 700cp for 3 moves, D 120 moves/EPD:C:\LittleBlitzer\2moves_80_100.epd(1749)
Time = 6636 sec elapsed, 0 sec remaining
 1.  SF_dev 8 threads         	549.0/1000	346-248-406  	(L: m=0 t=9 i=0 a=239)	(D: r=208 i=116 f=22 s=5 a=55)	(tpm=45.1 d=15.81 nps=10182858)
 2.  SF_dev 4 threads         	451.0/1000	248-346-406  	(L: m=1 t=0 i=0 a=345)	(D: r=208 i=116 f=22 s=5 a=55)	(tpm=45.1 d=15.57 nps=6991449)

46% faster
34.2 +/- 20 Elo points

Hyperthreaded is a whopping 46% faster and 34 Elo points stronger. Even more impressive is the SG NNUE (SV net used):

Code: Select all

Games Completed = 1000 of 1000 (Avg game length = 5.880 sec)
Settings = Gauntlet/128MB/1500ms+25ms/M 700cp for 3 moves, D 120 moves/EPD:C:\LittleBlitzer\2moves_80_100.epd(1749)
Time = 6402 sec elapsed, 0 sec remaining
 1.  SF NNUE 0633 8 threads                	582.0/1000	337-173-490  	(L: m=0 t=5 i=0 a=168)	(D: r=311 i=92 f=30 s=4 a=53)	(tpm=46.9 d=15.22 nps=7062790)
 2.  SF NNUE 0633 4 threads                  	418.0/1000	173-337-490  	(L: m=1 t=0 i=0 a=336)	(D: r=311 i=92 f=30 s=4 a=53)	(tpm=46.8 d=15.07 nps=4671940)

51% faster
57.5 +/- 19 Elo points.


Hyperthreaded is as much as 51% faster and 57 Elo points stronger.

I never before had such results using hyperthreading. What happens? The games are ultra-fast, but I see even better speed-up at longer TC (up to 60% speed-up with SF NNUE). Also, it seems that SF NNUE profits a lot from more threads.

What's that?

Did you check for time losses - I would have many times losses at that TC.

[MikeB]

running 8 threads vs 4 threads now, 8 concurrent games, CPU usage is running about 77% on a 32 core, 64 threads 3970x, a lot more games. to go
agree that it is better than expected - I would have guessed about 50 Elo better with my setup. Your results are little more amazing since it is only 50% increase in nps, whereas my speed differential is probably closer to 100% - since the resources are being shared more equitably. Still for SMP , very good results.

Code: Select all

PGN File: c:/cluster.mfb/pgn/08080049.pgn
Time Control: Time Control-> base+inc: 10+0.200
Games: 6000
Threads:
Hash: 256

Current date : time (EDST)
Date: 08/08/20 : 01:10:36

Projected-> Time: 8h:9m:44s
     Run -> Time: 0h:21m:27s

250 game(s) loaded
Rank Name        Rating   Δ     +    -     #     Σ    Σ%     W    L    D   W%    =%   OppR
---------------------------------------------------------------------------------------------------------

   1 SF-T8-1802   3529   0.0   27   27   250  146.0  58.4   78   36  136  31.2  54.4  3471
   2 SF-T4-1802   3471  58.5   27   27   250  104.0  41.6   36   78  136  14.4  54.4  3529
---------------------------------------------------------------------------------------------------------

  Δ = delta from the next higher rated opponent
  # = number of games played
  Σ = total score, 1 point for win, 1/2 point for draw

LOS:
            SF SF
SF-T8-1802     99
SF-T4-1802   0

250 game(s) loaded

loops/scheduled: 60/1500