Evaluating CPU cores for optimisation - page 14

 
Aleksey Vyazmikin:

Thanks for the tests. Got the table together - it turns out that there is very little difference between Ryzen 3700x and Ryzen 3800x


Just the other day i bought Ryzen 3700x and completely changed the memory to 32g, it looks like economical processor, but it gets really hot, heat dissipation is really bad. 3800 model has slightly higher clock speed but these 100-200 mhz are given by additional ~40w, not efficient, i would take hundred off 3700x to make it a bit cooler

 
Fast235:
Welcome back. Where's the hat?
 
Fast235:

Just the other day i bought Ryzen 3700x and completely changed the memory to 32g, it looks like economical processor, but it gets really hot, heat dissipation is really bad. 3800 model has slightly higher frequency but these 100-200 mhz are given by extra ~40w, not efficient, i would take hundred off 3700x and make it a bit cooler.

Then let's have tests, just to assess the stability of the indicators on different systems. Tests preferably on build 2190.

I have ordered 3600x for work - it would be interesting to feel Ryzen :)

 
Vladimir Baskakov:
Welcome back. Where is the hat?

trend change

Aleksey Vyazmikin:

Then let's have tests, just to assess the stability of performance on different systems. Tests preferably on build 2190.

I have ordered 3600x for work - it's interesting to feel Ryzen :)

I am just doing 32g of memory, have already done it for 16, in an hour I will lay out 2700+16, 3700x+16 and 3700x+32gb

add

tested my EA:

R2700- 1:35 time // Ryzen 3700x 1:12 ///3700x+32gb 1:10 (single thread)

R2700 1:45 //Ryzen 3700x 1:44 ///3700x+32gb 1:21 (16 characters)

R2700 7:38 //Ryzen3700x 5:39 // 3700x+32gb 5:01 (single thread)

option with 16gb memory at 3400, with 32gb it's 3733mhz

test #2 shows the emphasis was on memory, 16gb is not enough for 16 threads, tester used almost 16gb of virtual memory

 

2019.12.04 09:22:24.352 Core 2 pass 6 returned result 1001000.000000 in 0:00:55.729
2019.12.04 09:22:24.400 Core 4 pass 2 returned result 1001000.000000 in 0:00:55.783
2019.12.04 09:22:24.640 Core 3 pass 0 returned result 1001000.000000 in 0:00:56.022
2019.12.04 09:22:24.727 Core 1 pass 4 returned result 1001000.000000 in 0:00:56.110
2019.12.04 09:23:22.770 Core 2 pass 7 returned result 1001000.000000 in 0:00:58.415
2019.12.04 09:23:23.068 Core 4 pass 3 returned result 1001000.000000 in 0:00:58.663
2019.12.04 09:23:23.173 Core 3 pass 1 returned result 1001000.000000 in 0:00:58.528
2019.12.04 09:23:23.545 Core 1 pass 5 returned result 1001000.000000 in 0:00:58.813
2019.12.04 09:23:23.545 Tester optimization finished, total passes 8
2019.12.04 09:23:23.556 Statistics optimization done in 1 minutes 55 seconds
2019.12.04 09:23:23.556 Statistics shortest pass 0:00:55.729, longest pass 0:00:58.813, average pass 0:00:57.257
2019.12.04 09:23:23.556 Statistics 8000 frames (3.14 Mb total, 412 bytes per frame) received
2019.12.04 09:23:23.556 Statistics local 8 tasks (100%), remote 0 tasks (0%), cloud 0 tasks (0%)

2019.12.04 09:29:02.659 Core 1 Intel Core i7-4790K @ 4.00GHz, 16255 MB

 

2pc Intel® Xeon® Gold 6126, 4pc Samsung DDR4 2666 Registered ECC LRDIMM 64Gb

I'll let you know when I'm done with these gizmos

 
Fast235:

change of trend

I'm just doing 32g memory, I've already done 16, in an hour I'll post 2700+16, 3700x+16 and 3700x+32gb

add

tested my EA:

R2700- 1:35 time // Ryzen 3700x 1:12 ///3700x+32gb 1:10 (single thread)

R2700 1:45 //Ryzen 3700x 1:44 ///3700x+32gb 1:21 (16 characters)

R2700 7:38 //Ryzen3700x 5:39 // 3700x+32gb 5:01 (single thread)

option with 16gb memory at 3400, with 32gb it's 3733mpg

test #2 shows the emphasis was in memory, 16gb for 16 threads is very little, the tester used another virtual almost 16gb

Yes it's clear that here everything mainly rests in data transfer between drive and processor, it's possible to minimize it only by increasing RAM...

Judging by the results, floating point calculations are poorly involved, increment by improving coprocessor should be more. But, the information is interesting - maybe for usual EAs this is not necessary especially, I wonder...

 
SEM:

2019.12.04 09:22:24.352 Core 2 pass 6 returned result 1001000.000000 in 0:00:55.729
2019.12.04 09:22:24.400 Core 4 pass 2 returned result 1001000.000000 in 0:00:55.783
2019.12.04 09:22:24.640 Core 3 pass 0 returned result 1001000.000000 in 0:00:56.022
2019.12.04 09:22:24.727 Core 1 pass 4 returned result 1001000.000000 in 0:00:56.110
2019.12.04 09:23:22.770 Core 2 pass 7 returned result 1001000.000000 in 0:00:58.415
2019.12.04 09:23:23.068 Core 4 pass 3 returned result 1001000.000000 in 0:00:58.663
2019.12.04 09:23:23.173 Core 3 pass 1 returned result 1001000.000000 in 0:00:58.528
2019.12.04 09:23:23.545 Core 1 pass 5 returned result 1001000.000000 in 0:00:58.813
2019.12.04 09:23:23.545 Tester optimization finished, total passes 8
2019.12.04 09:23:23.556 Statistics optimization done in 1 minutes 55 seconds
2019.12.04 09:23:23.556 Statistics shortest pass 0:00:55.729, longest pass 0:00:58.813, average pass 0:00:57.257
2019.12.04 09:23:23.556 Statistics 8000 frames (3.14 Mb total, 412 bytes per frame) received
2019.12.04 09:23:23.556 Statistics local 8 tasks (100%), remote 0 tasks (0%), cloud 0 tasks (0%)

2019.12.04 09:29:02.659 Core 1 Intel Core i7-4790K @ 4.00GHz, 16255 MB

Thanks for the tests, but there are 3 EAs in the thread for which information is being collected. It is not clear which one you used for the tests. It is advisable to test on 3 EAs, according to the instructions - i.e. in"Mathematical calculations" mode.

 
Franklins100:

2pc Intel® Xeon® Gold 6126, 4pc Samsung DDR4 2666 Registered ECC LRDIMM 64Gb

I'll let you know when I run these gizmos.

Looking forward to it.

 
I do not understand why I changed the 16gb ram to 32, the tester as ate 16ram +16 virtual, now it eats 32ram +31 virtual, nonsense, I bought 32gb SSD to unload and prolong his life, and Pish. a day by terrabyte writes, poor ssd
Reason: