Evaluating CPU cores for optimisation - page 4

 
Georgiy Merts:

Above - Tree_Brut_TestPL_F

Here's the second one, in the same configuration:

Same (without F) when running 12 virtual cores:

Expert "with F" when running 12 virtual cores:

I'm going to turn off hypertrading now...

Expert "without F" and without hypertrading with six passes:

Expert "with F" and without hypertrading with six passes:

Very strange result - it turns outTree_Brut_TestPL without hypertrading disabled in bios 6 agents average pass 0:00:33.712, and with it disabled about 1:10 - no confusion?

While forTree_Brut_TestPL_F Expert Advisorwithout hypertradition disabled there are 6 agents" average pass 0:01:10.931"in the bios and "average pass 0:01:10.621" when disabled, i.e. there is no difference.

 

Next Subject Phenom II X4 960T 3000Mhz

2019.08.11 14:08:20.656 Terminal        Windows 7 Service Pack 1 (build 7601) x64, IE 11, AMD Phenom II X4 960 T Processor, Memory: 3182 / 4063 Mb, Disk: 57 / 92 Gb, GMT+3

Tree_Brut_TestPL - 4 agents

14:11:32.289    Core 3  pass 1 returned result 1001000.00 in 0:02:12.098
14:11:32.813    Core 2  pass 3 returned result 1001000.00 in 0:02:12.530
14:11:33.038    Core 1  pass 0 returned result 1001000.00 in 0:02:12.847
14:11:34.409    Core 4  pass 2 returned result 1001000.00 in 0:02:14.138
14:11:34.409    Tester  optimization finished, total passes 4
14:11:34.419    Statistics      optimization done in 2 minutes 15 seconds
14:11:34.419    Statistics      shortest pass 0:02:12.098, longest pass 0:02:14.138, average pass 0:02:12.903


Tree_Brut_TestPL_F - 4 agents

14:18:27.493    Core 4  pass 3 returned result 1001000.00 in 0:03:32.381
14:18:28.165    Core 1  pass 0 returned result 1001000.00 in 0:03:33.080
14:18:28.328    Core 2  pass 1 returned result 1001000.00 in 0:03:33.223
14:18:33.135    Core 3  pass 2 returned result 1001000.00 in 0:03:38.035
14:18:33.135    Tester  optimization finished, total passes 4
14:18:33.145    Statistics      optimization done in 3 minutes 40 seconds
14:18:33.145    Statistics      shortest pass 0:03:32.381, longest pass 0:03:38.035, average pass 0:03:34.179

 

Testing the FX-8350 again, but it's running at 4200Mhz and in one memory channel and another mother on the 760G chip

2019.08.11 08:54:57.931 Terminal        Windows 7 Service Pack 1 (build 7601) x64, IE 11, AMD FX-8350 Eight-Core Processor , Memory: 14341 / 16126 Mb, Disk: 106 / 272 Gb, GMT+3

Tree_Brut_TestPL - 4 agents

RN      0       09:02:08.810    Core 3  pass 2 returned result 1001000.00 in 0:01:08.306
EG      0       09:02:08.894    Core 2  pass 6 returned result 1001000.00 in 0:01:08.337
LL      0       09:02:08.987    Core 4  pass 4 returned result 1001000.00 in 0:01:08.427
LR      0       09:02:09.478    Core 1  pass 0 returned result 1001000.00 in 0:01:09.074
PK      0       09:03:16.943    Core 2  pass 7 returned result 1001000.00 in 0:01:08.057
NP      0       09:03:17.684    Core 3  pass 3 returned result 1001000.00 in 0:01:08.882
PF      0       09:03:17.794    Core 4  pass 5 returned result 1001000.00 in 0:01:08.816
NO      0       09:03:18.476    Core 1  pass 1 returned result 1001000.00 in 0:01:09.007
EJ      0       09:03:18.476    Tester  optimization finished, total passes 8
KN      0       09:03:18.486    Statistics      optimization done in 2 minutes 18 seconds
EQ      0       09:03:18.486    Statistics      shortest pass 0:01:08.057, longest pass 0:01:09.074, average pass 0:01:08.613

Tree_Brut_TestPL - 8 agents

KP      0       09:06:39.083    Core 1  pass 3 returned result 1001000.00 in 0:01:58.552
CI      0       09:06:40.856    Core 4  pass 0 returned result 1001000.00 in 0:02:00.327
FN      0       09:06:42.195    Core 3  pass 2 returned result 1001000.00 in 0:02:01.663
GD      0       09:06:43.023    Core 5  pass 6 returned result 1001000.00 in 0:02:01.890
RM      0       09:06:43.279    Core 2  pass 1 returned result 1001000.00 in 0:02:02.749
MS      0       09:06:43.675    Core 6  pass 7 returned result 1001000.00 in 0:02:02.395
KH      0       09:06:43.894    Core 8  pass 4 returned result 1001000.00 in 0:02:02.896
EQ      0       09:06:44.356    Core 7  pass 5 returned result 1001000.00 in 0:02:03.280
KH      0       09:06:44.357    Tester  optimization finished, total passes 8
FL      0       09:06:44.367    Statistics      optimization done in 2 minutes 04 seconds
CG      0       09:06:44.367    Statistics      shortest pass 0:01:58.552, longest pass 0:02:03.280, average pass 0:02:01.719

Tree_Brut_TestPL_F - 4 agents

PJ      0       09:20:56.349    Core 4  pass 2 returned result 1001000.00 in 0:01:47.739
PP      0       09:20:56.591    Core 2  pass 4 returned result 1001000.00 in 0:01:47.958
HI      0       09:20:57.103    Core 3  pass 0 returned result 1001000.00 in 0:01:48.680
HN      0       09:21:01.851    Core 1  pass 6 returned result 1001000.00 in 0:01:53.204
FD      0       09:22:44.340    Core 2  pass 5 returned result 1001000.00 in 0:01:47.763
HM      0       09:22:44.949    Core 4  pass 3 returned result 1001000.00 in 0:01:48.615
CR      0       09:22:45.221    Core 3  pass 1 returned result 1001000.00 in 0:01:48.132
FH      0       09:22:53.944    Core 1  pass 7 returned result 1001000.00 in 0:01:52.108
KN      0       09:22:53.944    Tester  optimization finished, total passes 8
IJ      0       09:22:53.954    Statistics      optimization done in 3 minutes 46 seconds
KM      0       09:22:53.954    Statistics      shortest pass 0:01:47.739, longest pass 0:01:53.204, average pass 0:01:49.274

Tree_Brut_TestPL_F - 8 agents

JO      0       09:11:41.826    Core 6  pass 0 returned result 1001000.00 in 0:03:23.474
QE      0       09:11:42.819    Core 5  pass 4 returned result 1001000.00 in 0:03:24.528
HJ      0       09:11:45.883    Core 2  pass 6 returned result 1001000.00 in 0:03:27.480
HS      0       09:11:45.900    Core 3  pass 7 returned result 1001000.00 in 0:03:27.437
FI      0       09:11:45.935    Core 4  pass 3 returned result 1001000.00 in 0:03:27.867
CN      0       09:11:47.761    Core 7  pass 1 returned result 1001000.00 in 0:03:29.327
ED      0       09:11:48.789    Core 8  pass 5 returned result 1001000.00 in 0:03:30.352
HM      0       09:11:51.431    Core 1  pass 2 returned result 1001000.00 in 0:03:33.033
CM      0       09:11:51.432    Tester  optimization finished, total passes 8
PI      0       09:11:51.442    Statistics      optimization done in 3 minutes 34 seconds
KP      0       09:11:51.442    Statistics      shortest pass 0:03:23.474, longest pass 0:03:33.033, average pass 0:03:27.937
 
Aleksey Vyazmikin:

Very strange result - it turns outTree_Brut_TestPL without switching off hypertrading in bios 6 agents average pass 0:00:33.712, and with switching off about 1:10 - no confusion?

While theTree_Brut_TestPL_F Expert Advisorwithout hypertradition disabled has 6 agents in the bios with" average pass 0:01:10.931" and with it disabled has "average pass 0:01:10.621", i.e. there is no difference.

I also think that hypertrading shouldn't make a difference when the task is purely computational. As I said before, the main benefit of virtual cores in optimization is precisely that we're not accessing memory, everything is in our cache. When I was testing different variants - I remember very well that on "real ticks" for two years there was no profit from hypertrading. And on the contrary, if I work with 1MOHLC I immediately get 10-20% profit. That is, the gain is due to caching.

The first result, on the other hand, surprises me...

 

The next test subject is a mother with two workstation/server CPUs, E5-2670 v1 processors at 3000Mhz (the mother forcibly boosts them)


2019.08.11 11:32:37.213 Terminal        Windows 7 Service Pack 1 (build 7601) x64, IE 11, Intel Xeon  E5-2670 0 @ 2.60 GHz, Memory: 60534 / 65483 Mb, Disk: 317 / 465 Gb, GMT+3

Tree_Brut_TestPL - 16 agents

RI      0       11:37:17.969    Core 001        pass 10 returned result 1001000.00 in 0:03:36.032
MP      0       11:37:25.419    Core 004        pass 4 returned result 1001000.00 in 0:03:43.584
KI      0       11:37:25.493    Core 007        pass 8 returned result 1001000.00 in 0:03:43.590
DL      0       11:37:28.716    Core 010        pass 16 returned result 1001000.00 in 0:03:46.730
JF      0       11:37:30.696    Core 011        pass 14 returned result 1001000.00 in 0:03:48.759
FI      0       11:37:34.878    Core 016        pass 26 returned result 1001000.00 in 0:03:52.774
QP      0       11:37:35.231    Core 006        pass 6 returned result 1001000.00 in 0:03:53.328
IE      0       11:37:37.138    Core 012        pass 24 returned result 1001000.00 in 0:03:55.033
IL      0       11:37:39.501    Core 015        pass 30 returned result 1001000.00 in 0:03:57.396
DE      0       11:37:40.320    Core 005        pass 2 returned result 1001000.00 in 0:03:58.658
GH      0       11:37:40.488    Core 014        pass 20 returned result 1001000.00 in 0:03:58.467
HQ      0       11:37:40.722    Core 003        pass 0 returned result 1001000.00 in 0:03:59.134
NE      0       11:37:41.021    Core 002        pass 28 returned result 1001000.00 in 0:03:58.915
FO      0       11:37:41.565    Core 008        pass 22 returned result 1001000.00 in 0:03:59.519
EQ      0       11:37:42.991    Core 009        pass 12 returned result 1001000.00 in 0:04:01.004
PH      0       11:37:45.825    Core 013        pass 18 returned result 1001000.00 in 0:04:03.805
FR      0       11:41:10.537    Core 001        pass 11 returned result 1001000.00 in 0:03:52.573
LG      0       11:41:17.444    Core 007        pass 9 returned result 1001000.00 in 0:03:51.955
MN      0       11:41:24.007    Core 011        pass 15 returned result 1001000.00 in 0:03:53.315
PR      0       11:41:30.690    Core 004        pass 5 returned result 1001000.00 in 0:04:05.275
GK      0       11:41:31.628    Core 010        pass 17 returned result 1001000.00 in 0:04:02.918
PM      0       11:41:33.720    Core 016        pass 27 returned result 1001000.00 in 0:03:58.847
DD      0       11:41:34.517    Core 015        pass 31 returned result 1001000.00 in 0:03:55.021
FN      0       11:41:35.195    Core 012        pass 25 returned result 1001000.00 in 0:03:58.063
IS      0       11:41:35.758    Core 006        pass 7 returned result 1001000.00 in 0:04:00.532
HJ      0       11:41:36.532    Core 002        pass 29 returned result 1001000.00 in 0:03:55.517
HM      0       11:41:38.036    Core 013        pass 19 returned result 1001000.00 in 0:03:52.217
PD      0       11:41:40.323    Core 003        pass 1 returned result 1001000.00 in 0:03:59.605
MI      0       11:41:41.865    Core 008        pass 23 returned result 1001000.00 in 0:04:00.305
QP      0       11:41:42.899    Core 014        pass 21 returned result 1001000.00 in 0:04:02.416
GI      0       11:41:42.992    Core 005        pass 3 returned result 1001000.00 in 0:04:02.676
IL      0       11:41:49.992    Core 009        pass 13 returned result 1001000.00 in 0:04:07.008
JI      0       11:41:49.992    Tester  optimization finished, total passes 32
NN      0       11:41:50.002    Statistics      optimization done in 8 minutes 09 seconds
DQ      0       11:41:50.002    Statistics      shortest pass 0:03:36.032, longest pass 0:04:07.008, average pass 0:03:56.092

Tree_Brut_TestPL - 32 agents

OO      0       12:21:37.757    Tester  optimization finished, total passes 32
KH      0       12:21:37.767    Statistics      optimization done in 5 minutes 16 seconds
CS      0       12:21:37.767    Statistics      shortest pass 0:05:07.092, longest pass 0:05:15.503, average pass 0:05:11.508

Tree_Brut_TestPL_F - 16 agents

MO      0       11:57:24.764    Tester  optimization finished, total passes 32
IK      0       11:57:24.774    Statistics      optimization done in 13 minutes 53 seconds
IS      0       11:57:24.774    Statistics      shortest pass 0:06:40.315, longest pass 0:07:04.229, average pass 0:06:50.431

Tree_Brut_TestPL_F - 32 agents

HJ      0       12:30:29.924    Tester  optimization finished, total passes 32
MM      0       12:30:29.934    Statistics      optimization done in 8 minutes 30 seconds
QD      0       12:30:29.934    Statistics      shortest pass 0:08:00.054, longest pass 0:08:28.112, average pass 0:08:13.088

I'm troubled by the scatter of agents in all variants - best - worst - up to 25 seconds - perhaps it is a payment for the dual-processor architecture, but I'm not sure - if anyone has a similar setup, please test it.

 
Georgiy Merts:

It seems to me, too, that hypertrading shouldn't make a difference when the task is purely computational. I've already said that the main benefit of virtual cores in optimization is that we don't access memory, everything is in the cache. When I was testing different variants - I remember very well that on "real ticks" for two years there was no profit from hypertrading. But if I work with 1MOHLC I immediately get 10-20% profit. That is the profit is due to caching.

But the first result - I'm surprised myself...

I'm not arguing with the theory, maybe it is so: all code with appendage in cache will fit in it and everything will be good and fast, while in the opposite case we will work with memory.

ConcerningTree_Brut_TestPL test- I think you must have made a mistake by switching off hypertrading in BIOS - try it again, the first Expert Advisor is faster than the second, so I assume this is a mistake.

 

So far, the ranking is based on the average run time in seconds - the two penultimate columns, and the last column shows the number of CPU runs per hour.

The table is filtered by the last column as the heaviest EA option in terms of computational resource consumption.



So far I'm surprised by poor result of E5-2670 processors, while it (2 of them) scored 2000 points in Cinabench r15, i7-8700 got only 1191 points! Something is not right here...

 
2019.08.11 18:16:13.866 Terminal        Windows 7 Service Pack 1 (build 7601) x64, IE 11, UAC, Intel Celeron  G3900 @ 2.80 GHz, Memory: 2473 / 3958 Mb, Disk: 63 / 111 Gb, GMT+3

Tree_Brut_TestPL - 2 agents

19:46.634	Core 1	pass 1 returned result 1001000.00 in 0:00:50.197
JL	0	18:19:46.701	Core 2	pass 0 returned result 1001000.00 in 0:00:50.267
RM	0	18:19:46.701	Tester	optimization finished, total passes 2
RI	0	18:19:46.711	Statistics	optimization done in 0 minutes 50 seconds
MM	0	18:19:46.711	Statistics	shortest pass 0:00:50.197, longest pass 0:00:50.267

Tree_Brut_TestPL_F - 2 agents

22:31.231       Core 1  pass 0 returned result 1001000.00 in 0:01:46.866
RO      0       18:22:38.982    Core 2  pass 1 returned result 1001000.00 in 0:01:54.603
LK      0       18:22:38.982    Tester  optimization finished, total passes 2
PO      0       18:22:38.992    Statistics      optimization done in 1 minutes 55 seconds
QK      0       18:22:38.992    Statistics      shortest pass 0:01:46.866, longest pass 0:01:54.603
 

2990WX test results

Tree_Brut_TestPL - 32 agents

2019.08.11 22:52:38.668 Tester  optimization finished, total passes 32
2019.08.11 22:52:38.678 Statistics      optimization done in 3 minutes 06 seconds
2019.08.11 22:52:38.678 Statistics      shortest pass 0:01:04.372, longest pass 0:02:09.139, average pass 0:01:42.144

Tree_Brut_TestPL - 64 agents

2019.08.11 22:46:49.456 Tester  optimization finished, total passes 64
2019.08.11 22:46:49.467 Statistics      optimization done in 3 minutes 16 seconds
2019.08.11 22:46:49.467 Statistics      shortest pass 0:02:41.837, longest pass 0:03:06.668, average pass 0:03:00.527

Tree_Brut_TestPL_F - 32 agents

2019.08.11 23:14:07.830 Tester  optimization finished, total passes 32
2019.08.11 23:14:07.840 Statistics      optimization done in 4 minutes 13 seconds
2019.08.11 23:14:07.840 Statistics      shortest pass 0:02:54.945, longest pass 0:04:12.540, average pass 0:03:37.639

Tree_Brut_TestPL_F - 64 agents

2019.08.11 23:07:06.525 Tester  optimization finished, total passes 64
2019.08.11 23:07:06.535 Statistics      optimization done in 5 minutes 52 seconds
2019.08.11 23:07:06.535 Statistics      shortest pass 0:05:31.519, longest pass 0:05:49.750, average pass 0:05:42.464

The results are very shocking and I cannot understand what's wrong - the compiler doesn't understand this architecture or something else - single runs are extremely mediocre and the spread is very large.

Yes, the overall result is the fastest for the processor, but I expected much better results! Who has Ryzen - let's test it!

 
Aleksey Vyazmikin:

2990WX test results

Tree_Brut_TestPL - 32 agents

Tree_Brut_TestPL - 64 agents

Tree_Brut_TestPL_F - 32 agents

Tree_Brut_TestPL_F - 64 agents

The results are very shocking and I cannot understand what's wrong - the compiler doesn't understand this architecture or something else - single runs are extremely mediocre and the spread is very large.

Yes, the overall result is the fastest for the processor, but I expected much better results! Who has Ryzen - let's test it!

The 2990WX is different. It consists of four Zeppelin crystals, with 32 processing cores. On the X399 platform, AMD has imposed some restrictions on this processor so that it doesn't hurt sales of EPYC server chips.

Chief among these limitations is the presence of only four memory controllers. Although there are two more Zeppelin chips, AMD calls them compute chips. This means they don't have access to local PCIe or DRAM, for that they have to address I/O components via Infinity Fabric. As there are twice as many crystals, the bandwidth of the Infinity fabric is twice as low, around 25Gb/s if DDR4-3200 memory is used.