Procena racunara

bmaxa:
Pa cl ti je tu najbitniji uzmes 3200/16 = 200 vs 2666/12 = 222, znaci oko 10% brzi na 2666
Kliknite za proširenje...
.Ne razumem kakva ti je to matematika//Gde si video da se to tako racuna? Prvi put to vidim...HAHHAHA ajd DOKAZI!! JA sam tebi izneo video gde se VIDI da to nije istina. Eto uzmi tvoju memoriju stavi na te vrednosti pusti TEST I prilozi. Samo to je recimo dokaz-sve ovo ostalo je trla baba lan..
 
ghost in the shell:
.Ne razumem kakva ti je to matematika//Gde si video da se to tako racuna? Prvi put to vidim...HAHHAHA ajd DOKAZI!! JA sam tebi izneo video gde se VIDI da to nije istina. Eto uzmi tvoju memoriju stavi na te vrednosti pusti TEST I prilozi. Samo to je recimo dokaz-sve ovo ostalo je trla baba lan..
Kliknite za proširenje...
Imam pametnija posla, radim ono za sta me placaju :P
Ja sam ti samo rekao, a ti veruj ili ne veruj ;)
Mislim lako mozes da proveris matematiku.
https://www.techpowerup.com/forums/threads/ddr4-frequency-vs-latency.232240/
Evo ovde malo drugacija matematika al isto dobijas 10% brze 2666 cl12 od 3200 cl6
 
bmaxa:
Imam pametnija posla, radim ono za sta me placaju :P
Ja sam ti samo rekao, a ti veruj ili ne veruj ;)
Mislim lako mozes da proveris matematiku.
https://www.techpowerup.com/forums/threads/ddr4-frequency-vs-latency.232240/
Evo ovde malo drugacija matematika al isto dobijas 10% brze 2666 cl12 od 3200 cl6
Kliknite za proširenje...
Joooj da izvini iimas 197 igara...al imas pametnija posla jeste...Joj sere mi se opd takvih kao ti koji su naj[pametniji za sve , kao...I programer, I gamer... I sve najbolje znas...
 
bmaxa:
Nemam 197, negde sad 170. Ali ti bezis pred cinjenicama. Gamer neki nisam, potrosim oko 15 minuta dnevno u proseku na igre...
Kliknite za proširenje...
imas nebitnbo, 170 igara-nije mali broj-a nisi gamer? Interesatno. P.C. Ne znam ko bezi pre cinjenicama a ja sam tebi pruzio DOKAZ znaci neko uzeo i pokazao kako, sta.. u PRAKSI! Bdw I na toim forumu nisu bas svi istog misljenja… Tako da-nisi, opet dokazao nista..
 
evo su rezultati dva testa test tlb memory latency i ramsp bw test.

Kod: 
~/Documents >>> cat memtest.txt                                                                                                                                                                          
2666/12

==================
~/projects/test-tlb >>> make run                                                                                                                                                             ±[●][master]
for i in 4k 8k 16k 32k 64k 128k 256k 512k 1M 2M 4M 6M 8M 16M 32M 64M 128M 256M ; do echo "$i:"; ./test-tlb -H $i 64; ./test-tlb $i 64 ; ./test-tlb -Hr $i 64; ./test-tlb -r $i 64; done
4k:
  1.12ns (~4.0 cycles)
  1.12ns (~4.0 cycles)
  1.12ns (~4.0 cycles)
  1.12ns (~4.0 cycles)
8k:
  1.12ns (~4.0 cycles)
  1.12ns (~4.0 cycles)
  1.12ns (~4.0 cycles)
  1.12ns (~4.0 cycles)
16k:
  1.13ns (~4.1 cycles)
  1.12ns (~4.0 cycles)
  1.12ns (~4.0 cycles)
  1.12ns (~4.0 cycles)
32k:
  1.17ns (~4.2 cycles)
  1.16ns (~4.2 cycles)
  1.15ns (~4.1 cycles)
  1.15ns (~4.1 cycles)
64k:
  1.75ns (~6.3 cycles)
  1.77ns (~6.4 cycles)
  3.37ns (~12.1 cycles)
  3.36ns (~12.1 cycles)
128k:
  1.79ns (~6.4 cycles)
  1.81ns (~6.5 cycles)
  3.36ns (~12.1 cycles)
  3.38ns (~12.2 cycles)
256k:
  1.77ns (~6.4 cycles)
  1.88ns (~6.8 cycles)
  3.36ns (~12.1 cycles)
  3.45ns (~12.4 cycles)
512k:
  1.84ns (~6.6 cycles)
  1.88ns (~6.8 cycles)
  3.44ns (~12.4 cycles)
  6.56ns (~23.6 cycles)
1M:
  1.85ns (~6.7 cycles)
  2.02ns (~7.3 cycles)
  8.26ns (~29.7 cycles)
  9.97ns (~35.9 cycles)
2M:
  1.84ns (~6.6 cycles)
  1.85ns (~6.6 cycles)
  9.08ns (~32.7 cycles)
 11.04ns (~39.8 cycles)
4M:
  1.85ns (~6.7 cycles)
  1.89ns (~6.8 cycles)
 10.02ns (~36.1 cycles)
 13.59ns (~48.9 cycles)
6M:
  2.07ns (~7.5 cycles)
  2.74ns (~9.9 cycles)
 14.13ns (~50.9 cycles)
 35.88ns (~129.2 cycles)
8M:
  3.33ns (~12.0 cycles)
  4.00ns (~14.4 cycles)
 47.99ns (~172.8 cycles)
 58.74ns (~211.5 cycles)
16M:
  3.68ns (~13.2 cycles)
  4.57ns (~16.4 cycles)
 73.19ns (~263.5 cycles)
 81.83ns (~294.6 cycles)
32M:
  4.15ns (~14.9 cycles)
  4.95ns (~17.8 cycles)
 80.04ns (~288.2 cycles)
 89.75ns (~323.1 cycles)
64M:
  3.62ns (~13.0 cycles)
  5.30ns (~19.1 cycles)
 82.89ns (~298.4 cycles)
 95.42ns (~343.5 cycles)
128M:
  3.87ns (~13.9 cycles)
  5.05ns (~18.2 cycles)
 85.96ns (~309.5 cycles)
 98.93ns (~356.2 cycles)
256M:
  4.05ns (~14.6 cycles)
  5.13ns (~18.5 cycles)
 90.22ns (~324.8 cycles)
102.80ns (~370.1 cycles)

~/bench/ramsmp-3.5.0 >>> ./ramsmp -b 12                                                                                                                                                                  
RAMspeed/SMP (Linux) v3.5.0 by Rhett M. Hollander and Paul V. Bolotoff, 2002-09

8Gb per pass mode, 2 processes

SSE       Copy:      22356.14 MB/s
SSE       Scale:     22174.35 MB/s
SSE       Add:       24781.86 MB/s
SSE       Triad:     24858.88 MB/s
---
SSE       AVERAGE:   23542.81 MB/s


3200/16
==============================

for i in 4k 8k 16k 32k 64k 128k 256k 512k 1M 2M 4M 6M 8M 16M 32M 64M 128M 256M ; do echo "$i:"; ./test-tlb -H $i 64; ./test-tlb $i 64 ; ./test-tlb -Hr $i 64; ./test-tlb -r $i 64; done
4k:
  1.11ns (~4.0 cycles)
  1.11ns (~4.0 cycles)
  1.11ns (~4.0 cycles)
  1.11ns (~4.0 cycles)
8k:
  1.11ns (~4.0 cycles)
  1.11ns (~4.0 cycles)
  1.11ns (~4.0 cycles)
  1.11ns (~4.0 cycles)
16k:
  1.11ns (~4.0 cycles)
  1.12ns (~4.0 cycles)
  1.11ns (~4.0 cycles)
  1.11ns (~4.0 cycles)
32k:
  1.16ns (~4.2 cycles)
  1.16ns (~4.2 cycles)
  1.14ns (~4.1 cycles)
  1.15ns (~4.1 cycles)
64k:
  1.81ns (~6.5 cycles)
  1.79ns (~6.5 cycles)
  3.36ns (~12.1 cycles)
  3.35ns (~12.1 cycles)
128k:
  1.77ns (~6.4 cycles)
  1.78ns (~6.4 cycles)
  3.36ns (~12.1 cycles)
  3.35ns (~12.1 cycles)
256k:
  1.78ns (~6.4 cycles)
  1.90ns (~6.8 cycles)
  3.35ns (~12.1 cycles)
  3.42ns (~12.3 cycles)
512k:
  1.74ns (~6.2 cycles)
  1.81ns (~6.5 cycles)
  3.41ns (~12.3 cycles)
  6.54ns (~23.5 cycles)
1M:
  1.81ns (~6.5 cycles)
  1.99ns (~7.2 cycles)
  8.18ns (~29.4 cycles)
 10.03ns (~36.1 cycles)
2M:
  1.88ns (~6.8 cycles)
  2.06ns (~7.4 cycles)
  9.00ns (~32.4 cycles)
 11.02ns (~39.7 cycles)
4M:
  1.89ns (~6.8 cycles)
  2.07ns (~7.4 cycles)
  9.41ns (~33.9 cycles)
 11.57ns (~41.7 cycles)
6M:
  1.88ns (~6.8 cycles)
  2.64ns (~9.5 cycles)
  9.56ns (~34.4 cycles)
 18.01ns (~64.8 cycles)
8M:
  3.10ns (~11.2 cycles)
  3.71ns (~13.4 cycles)
 15.05ns (~54.2 cycles)
 31.84ns (~114.6 cycles)
16M:
  3.54ns (~12.8 cycles)
  4.22ns (~15.2 cycles)
 64.87ns (~233.5 cycles)
 72.14ns (~259.7 cycles)
32M:
  3.53ns (~12.7 cycles)
  4.97ns (~17.9 cycles)
 74.47ns (~268.1 cycles)
 84.38ns (~303.8 cycles)
64M:
  3.54ns (~12.8 cycles)
  4.77ns (~17.2 cycles)
 79.44ns (~286.0 cycles)
 89.67ns (~322.8 cycles)
128M:
  3.54ns (~12.8 cycles)
  4.84ns (~17.4 cycles)
 81.92ns (~294.9 cycles)
 95.85ns (~345.1 cycles)
256M:
  3.54ns (~12.8 cycles)
  5.00ns (~18.0 cycles)
 83.46ns (~300.4 cycles)
 98.68ns (~355.3 cycles)

RAMspeed/SMP (Linux) v3.5.0 by Rhett M. Hollander and Paul V. Bolotoff, 2002-09

8Gb per pass mode, 2 processes

SSE       Copy:      27849.34 MB/s
SSE       Scale:     27792.54 MB/s
SSE       Add:       31894.16 MB/s
SSE       Triad:     31871.91 MB/s
---
SSE       AVERAGE:   29851.99 MB/s

RAMspeed/SMP (Linux) v3.5.0 by Rhett M. Hollander and Paul V. Bolotoff, 2002-09

8Gb per pass mode, 2 processes

3000/14
=============================
~/projects/test-tlb >>> make run                                                                                                                                                            ±[●●][master]
for i in 4k 8k 16k 32k 64k 128k 256k 512k 1M 2M 4M 6M 8M 16M 32M 64M 128M 256M ; do echo "$i:"; ./test-tlb -H $i 64; ./test-tlb $i 64 ; ./test-tlb -Hr $i 64; ./test-tlb -r $i 64; done
4k:
  1.11ns (~4.0 cycles)
  1.11ns (~4.0 cycles)
  1.11ns (~4.0 cycles)
  1.11ns (~4.0 cycles)
8k:
  1.12ns (~4.0 cycles)
  1.11ns (~4.0 cycles)
  1.11ns (~4.0 cycles)
  1.11ns (~4.0 cycles)
16k:
  1.11ns (~4.0 cycles)
  1.11ns (~4.0 cycles)
  1.11ns (~4.0 cycles)
  1.11ns (~4.0 cycles)
32k:
  1.16ns (~4.2 cycles)
  1.16ns (~4.2 cycles)
  1.15ns (~4.1 cycles)
  1.15ns (~4.1 cycles)
64k:
  1.77ns (~6.4 cycles)
  1.79ns (~6.4 cycles)
  3.36ns (~12.1 cycles)
  3.38ns (~12.2 cycles)
128k:
  1.77ns (~6.4 cycles)
  1.81ns (~6.5 cycles)
  3.36ns (~12.1 cycles)
  3.37ns (~12.1 cycles)
256k:
  1.78ns (~6.4 cycles)
  1.86ns (~6.7 cycles)
  3.35ns (~12.1 cycles)
  3.43ns (~12.4 cycles)
512k:
  1.80ns (~6.5 cycles)
  1.83ns (~6.6 cycles)
  3.46ns (~12.4 cycles)
  6.57ns (~23.6 cycles)
1M:
  1.84ns (~6.6 cycles)
  1.91ns (~6.9 cycles)
  8.23ns (~29.6 cycles)
 10.02ns (~36.1 cycles)
2M:
  1.77ns (~6.4 cycles)
  1.98ns (~7.1 cycles)
  9.05ns (~32.6 cycles)
 10.77ns (~38.8 cycles)
4M:
  1.87ns (~6.7 cycles)
  2.00ns (~7.2 cycles)
  9.48ns (~34.1 cycles)
 11.33ns (~40.8 cycles)
6M:
  1.91ns (~6.9 cycles)
  2.22ns (~8.0 cycles)
  9.66ns (~34.8 cycles)
 12.24ns (~44.1 cycles)
8M:
  3.15ns (~11.3 cycles)
  3.85ns (~13.9 cycles)
 16.49ns (~59.4 cycles)
 29.01ns (~104.4 cycles)
16M:
  3.62ns (~13.0 cycles)
  5.14ns (~18.5 cycles)
 66.15ns (~238.1 cycles)
 68.36ns (~246.1 cycles)
32M:
  3.53ns (~12.7 cycles)
  4.88ns (~17.6 cycles)
 75.77ns (~272.8 cycles)
 85.55ns (~308.0 cycles)
64M:
  3.44ns (~12.4 cycles)
  3.98ns (~14.3 cycles)
 80.76ns (~290.7 cycles)
 92.80ns (~334.1 cycles)
128M:
  3.41ns (~12.3 cycles)
  5.12ns (~18.4 cycles)
 83.24ns (~299.7 cycles)
 96.89ns (~348.8 cycles)
256M:
  3.58ns (~12.9 cycles)
  5.02ns (~18.1 cycles)
 84.06ns (~302.6 cycles)
100.41ns (~361.5 cycles)

~/bench/ramsmp-3.5.0 >>> ./ramsmp -b 12
RAMspeed/SMP (Linux) v3.5.0 by Rhett M. Hollander and Paul V. Bolotoff, 2002-09

8Gb per pass mode, 2 processes

SSE       Copy:      25839.81 MB/s
SSE       Scale:     25817.76 MB/s
SSE       Add:       29484.58 MB/s
SSE       Triad:     29668.62 MB/s
---
SSE       AVERAGE:   27702.69 MB/s

jedino bw pokazuje neku razliku, latencija tu ista, sto ce reci sve jedno.
 
U ovome zadnjem SSE average se vidi jasna razlika..I to uprednost 3200.
SSE AVERAGE: 23542.81 MB/s to je za 2666 CL 12
dalje 3200 CL 16 SSE AVERAGE: 29851.99 MB/s
3000 CL 14 SSE AVERAGE: 27702.69 MB/s
Ima jos nekih sitnjih razlika. za 256 Mb vreme pristupa je najmanje za 320098.68ns (~355.3 cycles), za 26666 cl12102.80ns (~370.1 cycles) -, itd.U sustini cak i da je isto-a nije-opet ne stoji tvoja prica koju non stop forsiras"bolje 2666 CL 12 nego 3200 CL 16" Pogotovo sto sam vec pokazao da od vecih frekvencija igre imaju solidan benefit.
 
Prijavite se ili registrujte da biste poslali poruku.

Back
Top