201

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

At 4 on 6  not including some other useful FAUGH type of asynchronous preloading passing a cache.
In studio, differently it is not considered productivity digit. On 8 above laid out, thanks big)
And you want to tell that this monster more slowly your lousy  from it 2 commands for clock period on a kernel
Yes, about it I also want to tell) in  too not a bast shoe a Russian cabbage soup sups, and "lousy" MIPS - it only in your imagination lousy. Can at the same time 4 with Chinese  .

202

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

(facebook.1486783941) Valeriy Shunkov
Yes, about it I also want to tell) in  too not a bast shoe a Russian cabbage soup sups, and "lousy" MIPS - it only in your imagination lousy. Can at the same time 4 with Chinese  .
Well it only in the parallel Universe the processor of 2 operations executing only for clock period on a kernel can faster appear the processor which about 25 operations for clock period:gigi are capable to execute:

203

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

(facebook.1486783941) Valeriy Shunkov
Well more slowly it  a lie, what for say lies?
There was a sampling in one institution of all competing domestic , you for certain have results...
Your chip and Baikal-T1 in spec 2006 to put it mildly not  unlike e2s...

204

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

(facebook.1486783941) Valeriy Shunkov
In studio, differently it is not considered productivity Digit.
There was here in this subject one person on Besson's surname. It launched on 4 the test which was the candidate on switching-on in SPEC 2006:
Athlon-64 2 GHz 8.63 14.8 10.1 9.34 = 42.9
Itanium-2 1.6 GHz 4.25 7.65 7.10 5.45 = 24.45
Core 2 3.0 GHz 2.93 5.52 3.64 3.52 = 15.61
Westmere 3.06 GHz 2.18 3.98 2.95 2.56 = 11.66
Elbrus-4S 0.72 GHz 5.13 10.9 12.7 6.38 = 35.12
In addition - results of scaling at start of four flows in comparison with one:
Westmere 3.06 GHz 2.18 3.98 2.95 2.56 = 11.66 (1 flow)
2.50 4.56 3.29 2.76 = 13.10 (4 flows)
87 % OF 87 % OF 89.7 % OF 93 % OF 89 %
Elbrus-4S 0.72 GHz 5.13 10.9 12.7 6.38 = 35.12 (1 flow)
5.9 11.75 13.3 6.8 = 37.7 (4 flows)
87 % OF 93 % OF 95.5 % OF 94 % OF 93 %
Westmere: Elbrus 2.35 2.74 4.31 2.49 3.01 (1 flow)
2.36 2.58 4.04 2.46 2.88 (4 flows)
===============
Results - execution time in seconds. Your Komdiv tears a navel in this test.
http://www.ipmnet.ru/~bess/bess-preprint2012.pdf
For 4 tests SPEC 2000 for a long time have been already published:
http://imgur.com/a/SC26F
It appears faster than Pentium M 1 GHz.
But it is old tests and the old compiler. On new lcc only at the expense of aggressive inlajn-substitution of functions on SPEC 2006 productivity increased by 30-40 %
And here with  such focuses not :gigi:

205

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

. Pentium-M and Atom D510 . Pentium MMX 233 did not quit in the field. i486DX2-66 affected the patient.:gigi:

206

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

Razboynik
For 4 tests SPEC 2000 for a long time have been already published:
It not 4, and  2
Adding from 12/19/2016 21:04:
Razboynik Multiklet.
Well generally at Multikleta only talks were about multi-cells.
In a reality were restricted only to 4-cages, what even it is less on number , than in normal processors.
Accordingly, any especially new ideas for implementation also it was not required. In particular, registers the general.
Now looked on a site - - is RTL a 16-cellular kernel. Basically, here already there should be any informative problems. But all the same it is not enough in comparison with the same GPU

207

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

matik :
vss
To the Customer is better to change nothing in operating system
Know, I would suggest to cease to consider a customer as the fool.

to convince a serious customer to pass to the new version of an operating system difficultly enough. Just because it not the fool.
Adding from 12/19/2016 22:01:

vindium :
Chineses persistently do not want to be fenced off from the remaining world by unique architecture.

yes very many people does not develop architecture. Whether it means, what development in this area are absolutely not necessary?

208

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

vss
Whether it Means, what development in this area are absolutely not necessary?
About! Yes it already almost academic question. Certainly it is necessary. And here to provide sale of the academic product with support of the device of state compulsion, without looking at direct and indirect expenses of this state - it is not necessary.

209

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

There was here in this subject one person on Besson's surname. It launched on 4 the test which was the candidate on switching-on in SPEC 2006:
Well as, it turns out 4 "on the order faster" ? In my opinion on the order it is impossible.
But it is anyway good that started digits to consider, instead of mythology)

210

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

alexsr
Again poor jugglings and lie? :-\:down::mad:
INTEL PENTIUM M 780, 1.00 GHz
SPEC CPU 2006: 5.04
SPEC FPU 2006: 3.69
INTEL CORE2DUO E6850, 1.00 GHz
SPEC CPU 2006: 7.1
SPEC FPU 2006: 6.7

What for processors such? Where such to buy? At Intel such is not present.:down:
And results normal Pentium M 780 "for some reason" others:
SPEC CPU 2006 int - 11.4
SPEC CPU 2006 fp - 8.35
Core 2 Duo E6850:
SPEC CPU 2006 int - 22.6
SPEC CPU 2006 fp - 20.3
Again lie and jugglings:down::mad:
Why these people ALWAYS lie?:mad:

211

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

matik
Why these people ALWAYS lie?
INTEL PENTIUM M 780, 1.00 GHz
SPEC CPU 2006: 5.04
SPEC FPU 2006: 3.69
INTEL CORE2DUO E6850, 1.00 GHz
SPEC CPU 2006: 7.1
SPEC FPU 2006: 6.7
Similar in a post results for the processor with  so it is convenient to compare how many calculations of architecture of processors are resulted do for one clock period.

212

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

(vkontakte.56442331) Alexander Kobjakov how many calculations of architecture of processors do for one clock period.
As SPEC the complex test of the computer (including at least the processor, a subsystem of storage and the compiler) to normalize its results on frequency - - senseless enough action.

213

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

(vkontakte.56442331) Alexander Kobjakov
Similar in a post results for the processor with  are resulted
It is traditional juggling mad
They compare not real results, and "normed on frequency". And all why? All because their microprocessors do not reach frequencies of processors of Intel.
Only one of methods "to smooth impression" :-\
So it is convenient to compare how many calculations of architecture of processors do for one clock period.
It is absolutely senseless to compare "how many calculations for clock period". At VLIW processors and at normal all miscellaneous: both clock period, and achievable frequency, and all remaining.
Calculation for clock period "is no point in parameter", it is initially known that clock period at VLIW processors "heavier".
Adding from 12/20/2016 00:45:
The Message named unsuccessful: alexsr
But a traditional hysterics when caught on lies