26

Re: Domestic microprocessors (2) (a part 8)

milaro222 :

B4rb4ru5 :
I in the sense that Russian manual, as a rule, am able to focus resources on the decision of essential tasks, let and to the detriment of needs of civilians. Why they how it normally happens, did not select  and did not squeeze out of them processor variations of Iskander and -400? Or designing of the microprocessor demands the full creative freedom and is possible only on inspiration, by the way, in garage?

And than you does not arrange Elbrus as a processor variation of Iskander and -400, Iskander and -400 it on the technological level of system far not advanced and cheap, with the concept and element basis 10-20 summer prescription, but powerful enough for parrying of threats of more developed armies.

Want to tell, in the technical plan backward, but in them gunpowder suffices?)

Elbrus, it also the cheap decision which at minimum expenses provides necessary needs, and at due level of implementation is possible and in a state to compete to the modern development.

That is, if to implement Elbrus in modern technical process, it can make a competition of production of Intel, AMD?

27

Re: Domestic microprocessors (2) (a part 8)

And meanwhile in :

Aggressive inlajn-substitution of functions for VLIW-architecture
The offered algorithm has been implemented in the industrial optimizing compiler for architecture Elbrus [16,17]. For empirical selection
Coefficients in functions of estimation Te and Tc tasks from packet SPEC CPU2000 [18] were used. Coefficients stole up so that
To minimize root mean square deviation of estimate and real values Te and Tc. For check of efficiency of the offered algorithm
29 tasks from packet SPEC CPU2006 [18] were used.
Samplings of time of execution of tasks were led on  the microprocessor
Elbrus 4C with clock frequency of 800 MHz. In a module-by-module mode of the assembly without the dynamic profile information (an option-O3-ffast)
For the account of inlajn-substitution execution time on the average decreased for 29 % , thus compile time increased all by 13 %, and the size of a binary file increased by 9.1 % (see table 1). In a mode "all program" with profile usage (options-O3-ffast-fwhole-fprofile-use) results turned out even more impressive: execution was accelerated on 41 % , compilation was decelerated on 12 %, the size of a binary file increased by 7.7 % on the average (see table 2). the Greatest acceleration has been reached on the task 447.dealII which was accelerated in 7.3 time at insignificant deceleration of compilation (0.9 %) and code size reduction (1.5 %).
http://www.ispras.ru/proceedings/docs/2 … _6_189.pdf

Time of execution of test SPEC 2006 to Elbrus-4S decreased on the average for 41 % only at the expense of aggressive inlajn-substitution of functions.

28

Re: Domestic microprocessors (2) (a part 8)

Well one more long-term dispute of this branch here dared - whether - Has  copy SPEC2006 and whether holds testing of the processors with its help. wink
And again that is characteristic - - results which could be compared to other processors - - we did not see. sad

29

Re: Domestic microprocessors (2) (a part 8)

VLev :
Copy SPEC2006 and whether is held by testing of the processors with its help. wink
And again that is characteristic - - results which could be compared to other processors - - we did not see. sad

We saw that tests test not processors, and a sheaf +.

30

Re: Domestic microprocessors (2) (a part 8)

vss
We saw that tests test not processors, and a sheaf +.
You about it learned only today?:rotate:

31

Re: Domestic microprocessors (2) (a part 8)

32

Re: Domestic microprocessors (2) (a part 8)

VLev :

vss : We saw that tests test not processors, and a sheaf +.

Not we, but you (). And for us it not a secret

Then in what sense of a phrase "results which could be compared to other processors"?

33

Re: Domestic microprocessors (2) (a part 8)

:insane:
system's processor, memory subsystem and compiler
-- It a minimum dial-up that it is possible to name the computer.
Without something one all talks about "processor" have academic character.
Such spherical horse in vacuum.
Adding from 12/8/2016 15:59:
Therefore results in SPEC are necessary - is the first step of the processor from "amusing " () in real life.
And this step from Elbrus meanwhile it is not visible sad

34

Re: Domestic microprocessors (2) (a part 8)

Here the same selection about any weapon of our bewitched friends.
https://www.youtube.com/watch?v=ELwjpNdcmLw
Question, as usual, one: well also what?

35

Re: Domestic microprocessors (2) (a part 8)

B4rb4ru5 :
Want to tell, in the technical plan backward, but in them gunpowder suffices?)

It is impossible to tell that they backward, simply use cheap decisions, instead of AFAR , instead of rockets with the active induction the semiactive.
a complex modern but its concept again to consist that if on something to put high-precision system of induction, it there should be a huge long loaf what in vain not to spend expensive electronics.

B4rb4ru5 :
That is, if to implement Elbrus in modern technical process, it can make a competition of production of Intel, AMD?

the Problem any more in technical process though he certainly plays a role, but in implementation of topology, a chip are projected on the basis of standard cells which do not allow to achieve high frequencies, and development  chips demands 10 times bigger means.
Anyway so explain the lag in .
For example SPARC V9 from  has frequency of 1 Ggts, precisely such Spark from Fudzhitsu on that technical process how much I remember, has frequency of 2.3 Ggts.

36

Re: Domestic microprocessors (2) (a part 8)

1. The roller was laid out "SHAKED IL", by the known Jewish russophob from the Jewish forum "War online".
2. And where on video it is written, what it is Russia and in 2016? I here see that it is the Ukrainian air defenses tried to launch the rocket with the delayed periods of maintenance.
3. On a Maidan highly skipped, "American"?:laugh:

37

Re: Domestic microprocessors (2) (a part 8)

[off] Concept "russophob" expresses the concept of the offended nation and (considering  imperial consciousness) conducts direct road to nazism. Everything, since the use in a branch several posts above, it is possible to erase, as senseless dispute with latent fascists. Words here are not convincing. [/off]
Adding from 12/9/2016 09:28:
[off] the Problem that logs in the eye do not note. It like a photographic lens which accurately represents that, on what it is focused, but not the internal construction. Defects become noticeable through other sensations - when an eye . [/off]

38

Re: Domestic microprocessors (2) (a part 8)

BlazeBlaze :
You will laugh, but expensive and difficult  appeared on MiG-31 because of impossibility to implement cheap rockets with the active induction. So peep about processors, instead of about

If it was so, all over the world already for a long time refused from expensive and difficult  and the more so , on that that with rockets with the active induction at anybody for a long time already problems are not present.
Adding from 12/9/2016 13:08:

BlazeBlaze :
You not , and invent. Still there photos of the tell-tale not in the Ukrainian form are laid out. There were explanations that these setting are received from  with storage in disgusting conditions.

The photo there doubtful enough, resembles at all a photoshop, and on paint.

39

Re: Domestic microprocessors (2) (a part 8)

milaro222
The Problem any more in technical process though he certainly plays a role, but in implementation of topology, a chip are projected on the basis of standard cells which do not allow to achieve high frequencies, and development  chips demands 10 times bigger means.
My God, how many time repeated a pattern... Please, cease to duplicate this bosh about standard cells and  circuits. All  circuits become on libraries of standard numeral cells, I as practising developer ASIC responsibly declare it to you.
Here still esteem how it becomes in NXP if want: http://www.mes-conference.ru/data/year2016/pdf/D011.pdf
In article results of data analysis on usage of libraries of standard cells from different suppliers and for different technologies in the real projects representing wide enough spectrum of a various class of GSI are presented:  controlers for industrial applications, microcontrollers for "the Internet of things (IoT)", high-efficiency decisions for wireless technologies, high speed interfaces, control diagrams for consumer electronics (games consoles, mobile audio), etc.

40

Re: Domestic microprocessors (2) (a part 8)

(facebook.1486783941) Valeriy Shunkov :
How it becomes in NXP,

And how it becomes in Intel?

41

Re: Domestic microprocessors (2) (a part 8)

(facebook.1486783941) Valeriy Shunkov :
My God, how many time repeated a pattern... Please, cease to duplicate this bosh about standard cells and  circuits. All  circuits become on libraries of standard numeral cells, I as practising developer ASIC responsibly declare it to you.

Well why bosh, same not I speak, so tell in , where that enough practising developers.
Anybody does not state that in Intel do each new processor from zero up to the last transistor, but standard cells at Intel not those standard cells which use it in , in Intel their study and optimization can be much better.
And I already resulted result, Spark  1, Spark Fudzhitsu of 2.3 GHz

42

Re: Domestic microprocessors (2) (a part 8)

vss
And how it becomes in Intel?
In the same way, synthesis of topology from the code in high level language through libraries of standard elements.
http://csg.csail.mit.edu/6.375/6_375_20 … -itj04.pdf
Forgive that the link old, but Intel at all not most open office.
milaro222
Well why bosh, same not I speak
And I also speak: "not duplicate bosh". On a fence practising developers too much that writes, especially when it is necessary to justify somehow that so much years transited, and things are right where they started. I to you gave confirming to the words of the reference to operations of the largest players of the world market, instead of people which promised to beat these players still twenty years ago, and all in any way even SPEC2006 cannot show.
But standard cells at Intel not those standard cells which use it in , in Intel their study and optimization can be much better.
The library consisted of over 1600 cells, covering over 130 unique logic functions, and associated collateral included over 20 file types. Most cells had variants implemented with different threshold voltage transistors to enable trade-offs between delay and leakage power.
Certainly, at Intel of library it is much better, but these are still standard cells, instead of mythical "completely  design" on which absence you eat to write off limitation of "Elbrus".
And I already resulted result, Spark  1, Spark Fudzhitsu of 2.3 GHz
The option "at someone curve hands" was not considered?

43

Re: Domestic microprocessors (2) (a part 8)

And titles good - Elbrus, Baikal, the Meadow-sweet...

44

Re: Domestic microprocessors (2) (a part 8)

(facebook.1486783941) Valeriy Shunkov :
milaro222
And I already resulted result, Spark  1, Spark Fudzhitsu of 2.3 GHz
The option "at someone curve hands" was not considered?

As it is surprising, here this option was not considered. Here concept VLIW and original architecture "Elbrus", instead of how this architecture is implemented within the limits of the specific Russian organization was criticized.

45

Re: Domestic microprocessors (2) (a part 8)

milaro222 And I already resulted result, Spark  1, Spark Fudzhitsu of 2.3 GHz
Valeriy Shunkov the Option "at someone curve hands" was not considered?
Judging by to characteristics R1000 , in particular - -
Power consumption 15 W
Number of kernels 4

Purpose  was to create , instead of the high-frequency processor (though for our realities 1 - - quite to itself high-frequency).
Anyway - - at Fudzhitsu even the one-nuclear version in times consumed more
Adding from 12/9/2016 21:46:
vss instead of how this architecture is implemented within the limits of the specific Russian organization.
Generally, the main criticism concerned realities to "the specific Russian organization".
"" generally became the nominal term.

46

Re: Domestic microprocessors (2) (a part 8)

Large, large projects! Forty  one messengers! (

47

Re: Domestic microprocessors (2) (a part 8)

In Russia develop , will use all that is, especially Elbruses, the purpose, to occupy the market in volume of 2.1 trillion dollars.
http://safe.cnews.ru/news/top/2016-12-0 … a_elbrusah

48

Re: Domestic microprocessors (2) (a part 8)

milaro222
Also there will be it so, open the safe in it Elbrus...

49

Re: Domestic microprocessors (2) (a part 8)

Dusty :
Also there will be it so, open the safe in it Elbrus...

And now unless not so? I will not be surprised that at sale of Elbrus there is a requirement that it was stored and exploited in a protected location with the restricted tolerance. It is almost assured that so it and is.

50

Re: Domestic microprocessors (2) (a part 8)

VLev :
Generally, the main criticism concerned realities to "the specific Russian organization".

Unless? On which errors in implementation of the architecture developed by Babayan, here it was specified, what lacks of the organization of the operations leading to errors, were considered.
Envy to is visible only what project has been carried out. It is not in full carried out, but it is rare when all is possible completely.