Question: When a CPU can perform a multiplication in 12 nanoseconds (ns), an addition in 1 ns,
and a subtraction in 1.5 ns, which of the following is the minimum CPU time, in
nanoseconds, for the calculation of “a×a – b×b ” ?
Answer: 14.5
I believe it optimizes the equation to (a-b)*(a+b) so it's subtraction + addition + multiplication = 12 + 1 + 1.5 = 14.5.
Though my math isn't the best around so if I'm wrong just comment and don't downvote so I can delete :D
Related
There are multiple queries of the form
Q(n,m) = (nC1*mC1) + (nC2*mC2) + (nC3*mC3) ... (nCk*mCk) where
k=min(n,m)
How to find the value of Q(n,m) in O(1) time complexity.
I tried pre-computing ncr[N][N] matrix and dp[N][N][N] where dp[n][m][min(n,m)] = Q(n,m).
This pre-computation takes O(N^3) time and queries can be answered in O(1) time now. But I'm looking for an approach in which pre-computation shouldn't take more O(N^2) time.
Solution for starting from C(n,0)*C(m,0) seems pretty simple
Q0(n,m) = C(n+m, m)
So for your formulation just subtract 1
Q(n,m) = C(n+m, m) - 1
Example: n=9, m=5
Dot product of 9-th and 5-th rows of Pascal's triangle is
1 9 36 84 126 126 84 36 9 1
1 5 10 10 5 1
1 + 45 + 360 + 840 + 630 + 126 = 2002 = C(14,5)
It might be proved with math induction starting from Q(n,1) but expressions are rather long.
I have discovered a truly marvelous demonstration of this proposition that this margin is too narrow to contain © Fermat ;)
I was stuck when trying to calculate for the speedup. So the question given was:
Question 1
If 50% of a program is enhanced by 2 times and the rest 50% is enhanced by 4 times then what is the overall speedup due to the enhancements? Hints: Consider that the execution time of program in the machine before enhancement (without enhancement) is T. Then find the total execution time after the enhancements, T'. The speedup is T/T'.
The only thing I know is speedup = execution time before enhancement/execution time after enhancement. So can I assume the answer is:
Speedup = T/((50/100x1/2) + (50/100x1/4))
Total execution time after the enhancement = T + speedup
(50/100x1/2) because 50% was enhanced by 2 times and same goes to the 4 times.
Question 2
Let us hypothetically imagine that the execution of (2/3)rd of a program could be made to run infinitely fast by some kind of improvement/enhancement in the design of a processor. Then how many times the enhanced processor will run faster compared with the un-enhanced (original) machine?
Can I assume that it is 150 times faster as 100/(2/3) = 150
Any ideas? Thanks in advance.
Let's start with question 1.
The total time is the sum of the times for the two halves:
T = T1 + T2
Then, T1 is enhanced by a factor of two. T2 is improved by a factor of 4:
T' = T1' + T2'
= T1 / 2 + T2 / 4
We know that both T1 and T2 are 50% of T. So:
T' = 0.5 * T / 2 + 0.5 * T / 4
= 1/4 * T + 1/8 * T
= 3/8 * T
The speed-up is
T / T' = T / (3/8 T) = 8/3
Question two can be solved similarly:
T' = T1' + T2'
T1' is reduced to 0. T2 is the remaining 1/3 of T.
T' = 1/3 T
The speed-up is
T / T' = 3
Hence, the program is three times as fast as before (or two times faster).
So I was told to ask this on here instead of StackExchage:
If I have a program P, which runs on a 2GHz machine M in 30seconds and is optimized by replacing all instances of 'raise to the power 4' with 3 instructions of multiplying x by. This optimized program will be P'. The CPI of multiplication is 2 and CPI of power is 12. If there are 10^9 such operations optimized, what is the percent of total execution time improved?
Here is what I've deduced so far.
For P, we have:
time (30s)
CPI: 12
Frequency (2GHz)
For P', we have:
CPI (6) [2*3]
Frequency (2GHz)
So I need to figure our how to calculate the time of P' in order to compare the times. But I have no idea how to achieve this. Could someone please help me out?
Program P, which runs on a 2GHz machine M in 30 seconds and is optimized by replacing all instances of 'raise to the power 4' with 3 instructions of multiplying x by. This optimized program will be P'. The CPI of multiplication is 2 and CPI of power is 12. If there are 10^9 such operations optimized,
From this information we can compute time needed to execute all POWER4 ("raise to the power 4) instructions, we have total count of such instructions (all POWER4 was replaced, count is 10^9 or 1 G). Every POWER4 instruction needs 12 clock cycles (CPI = clock per instruction), so all POWER4 were executed in 1G * 12 = 12G cycles.
2GHz machine has 2G cycles per second, and there are 30 seconds of execution. Total P program execution is 2G*30 = 60 G cycles (60 * 10^9). We can conclude that P program has some other instructions. We don't know what instructions, how many executions they have and there is no information about their mean CPI. But we know that time needed to execute other instructions is 60 G - 12 G = 48 G (total program running time minus POWER4 running time - true for simple processors). There is some X executed instructions with Y mean CPI, so X*Y = 48 G.
So, total cycles executed for the program P is
Freq * seconds = POWER4_count * POWER4_CPI + OTHER_count * OTHER_mean_CPI
2G * 30 = 1G * 12 + X*Y
Or total running time for P:
30s = (1G * 12 + X*Y) / 2GHz
what is the percent of total execution time improved?
After replacing 1G POWER4 operations with 3 times more MUL instructions (multiply by) we have 3G MUL operations, and cycles needed for them is now CPI * count, where MUL CPI is 2: 2*3G = 6G cycles. X*Y part of P' was unchanged, and we can solve the problem.
P' time in seconds = ( MUL_count * MUL_CPI + OTHER_count * OTHER_mean_CPI ) / Frequency
P' time = (3G*2 + X*Y) / 2GHz
Improvement is not so big as can be excepted, because POWER4 instructions in P takes only some part of running time: 12G/60G; and optimization converted 12G to 6G, without changing remaining 48 G cycles part. By halving only some part of time we get not half of time.
How can I calculate a floating point multiplicand in Verilog? So far, I usually use shift << 1024 , then floating point number become to integer. Then I do some operations, then >> 1024 to obtain a fraction again.
For example 0.3545 = 2^-2 + 2^-4 + ...
I have question about another way, like this. I don't know where does the minus (-) comes from:
0.46194 = 2^-1 - 2^-5 - 2^-7 + 2^-10.
I have just look this from someone. but as you way, that is represented like this
0.46194 = 2^-2 + 2^-3 + 2^-4 + 2^-6 + 2^-7 + 2^-10 + .... .
I don't understand how does it know the minus is used it?
How do we know when the minus needed to it? Also how can I apply to verilog RTL?
UPDATE : I understand the concept the using minus in operation. But Is there any other way to equation or methodologies what to make reduce expression what multiplying with power of 2?
UPDATE : how can we use this method in verilog? for example, I have leaned 0.46194 = 2^-1 - 2^-5 - 2^-7 + 2^-10. then this code was written like this in verilog. 0.011101101 ='hED = 'd237. So the point of the question is how can we apply it to application in verilog?
UPDATE : Sir Would you please check this one? there are a little difference result.
0.46194 = 0.011101101. I just tried like this
0.011101101
0.100T10T01
= 2^-1 - 2^-4 + 2^-5 - 2^-7 + 2^-9. = 0.462890625
Something different. What do I wrong?
Multiplication of a variable by a constant is often implemented by adding the variable to shifted versions of itself. This is much cheaper to put on an FPGA than a multiplier circuit accepting two variables.
You can get further savings when there's a sequence of 1-bits in the constant, by using subtraction as well. (A subtraction circuit is only equally expensive as addition.)
Consider the number 30 = 11110. It's equal to 16 + 8 + 4 + 2, but it's also equal to 32 - 2.
In general, a sequence of multiplicand 1-bits, or the sum of several successive powers of two, can be formed by adding the first power of two after the most significant bit, and subtracting the least significant bit. Hence, instead of 16x + ... + 2x, use 32x - 2x.
It doesn't matter if the sequence of 1-bits is part of a fraction or an integer. You're just applying the identity 2^a = 1 + ∑2^0 ... 2^(a-1), in other worsd ∑2^0 ... 2^a = 2^(a+1) - 1.
In a 4 bit base 2 number can have these values:
Base 2: Unsigned 4 bit integer,
2^3 2^2 2^1 2^0
8 4 2 1
If we have a 0111 it represents 7. If we were to multiply by this number using a shift add architecture it would take 3 clockcycles (3 shift and adds).
An optimisation to this is called CSD (Canonical Signed Digit. It allows minus one to be present in the 'binary numbers'. We shall represent -1 as one bar, or T as that looks like a one with a bar over the top.
100T represents 8 - 1 which is the same as 0111. It can be observed that long runs of 1's can be replaced with a the 0 that ends the run becoming 1 and the first 1 of the run becoming a -1, (T).
An example of conversion:
00111101111
01000T1000T
But if passed in two section we would get :
00111101111
0011111000T
010000T000T
We have taken a number that would take 8 clock cycles or 8 blocks of logic to compute and turned it into 3.
Related questions to fixed point values in Verilog x precision binary fixed point representation? and verilog-floating-points-multiplication.
To cover the follow up question:
To answer the follow up section about your question on CSD conversion. I will look at them as pure integers to simplify the numbers, this is the same as multiplying the values by 2^9 (9 fractional bits).
256 128 64 32 16 8 4 2 1
0 1 1 1 0 1 1 0 1
128 + 64 +32 + 8 +4 +1 => 237
Now with your CSD conversion:
256 128 64 32 16 8 4 2 1
1 0 0 T 1 0 T 0 1
256 -32 + 16 - 4 + 1 => 237
You can see your conversion was correct. I get 237* 2^-9 as 0.462890625, which matches your answer when converted back to fractional. The 0.46194 that you started with must have been a rounded version, or when quantised to 9 fractional bits gets truncated. This error is known as quantisation error. The most important thing here though is that you got the CSD conversion correct.
Suppose that you time a program as a function of N and produce
the following table.
N seconds
-------------------
19683 0.00
59049 0.00
177147 0.01
531441 0.08
1594323 0.44
4782969 2.46
14348907 13.58
43046721 74.99
129140163 414.20
387420489 2287.85
Estimate the order of growth of the running time as a function of N.
Assume that the running time obeys a power law T(N) ~ a N^b. For your
answer, enter the constant b. Your answer will be marked as correct
if it is within 1% of the target answer - we recommend using
two digits after the decimal separator, e.g., 2.34.
Can someone explain how to calculate this?
Well, it is a simple mathematical problem.
I : a*387420489^b = 2287.85 -> a = 387420489^b/2287.85
II: a*43046721^b = 74.99 -> a = 43046721^b/74.99
III: (I and II)-> 387420489^b/2287.85 = 43046721^b/74.99 ->
-> http://www.purplemath.com/modules/solvexpo2.htm
Use logarithms to solve.
1.You should calculate the ratio of the growth change from one row to the one next
N seconds
--------------------
14348907 13.58
43046721 74.99
129140163 414.2
387420489 2287.85
2.Calculate the change's ratio for N
43046721 / 14348907 = 3
129140163 / 43046721 = 3
therefore the rate of change for N is 3.
3.Calculate the change's ratio for seconds
74.99 / 13.58 = 5.52
Now let check the ratio between one more pare of rows to be sure
414.2 / 74.99 = 5.52
so the change's ratio for seconds is 5.52
4.Build the following equitation
3^b = 5.52
b = 1.55
Finally we get that the order of growth of the running time is 1.55.