i3 4130 3.4Ghz Memory and testing - Pentium and i5 Comparison results

[DonkeyOatie]

In the first thread, data was collected for the G3258 at various clock speeds, memory configurations, and with or without a discrete GPU. A similar set of data was then collected for the i5 4690K. This data is a comparison of those two sets with the i3 4130.


There are three data sets presented below. The first compares the i3 4130 with the G3258, the second compares the i5 4690K with the G3258, and the third compares the i5 4690K with the i3 4130.

The numbers are % difference in speed with the G3258 overclocked to the same 3.4 Ghz base clock and the i5 4690K underclocked to the same 3.4Ghz as the i3 4130.

The i5 4690K has Turbo-boost up to 3.9Ghz which the other two ships do not. It has been left activated to reflect 'real' results.

Everything else is the same. Memory is 2 x 4Gb at 1600Mhz.

1. GP use PCMark 7 = + 9.46%, + 17.70%, + 7.53% PCMark 8 C = + 6.84%, + 25.83%, + 17.77%, PCMark 8 A = + 8.32%, + 27.24%, + 17.46%.

2. Gaming, IS = +37.17%, + 75.74%, + 28.13%, CG = + 45.24%, + 90.02%, + 30.83%, SD = + 35.45%, + 91.42%, + 41.32%, FS = + 67.28%, + 103.96, + 21.92%, UH Low + 49.84%, + 62.62%, + 8.52%, Ultra + 76.64%, + 100.93%, + 13.76%, UV Low + 39.40%, + 76.63%, + 26.71%, Ultra + 60.00%, + 93.53%, + 20.96%.

3. Numeric 1 = - 0.67%, + 18.33%, + 19.12%, 2 = - 8.14%, + 19.26%, + 29.83%, 3 = + 4.54%, +61.25%, + 54.25%, 4 = + 14.87%, + 83.83%, + 60.03%, 5 = + 8.28%, + 63.60%, + 51.08%, 6 = + 11.60%, + 70.16%, + 52.47%, 7 = + 16.93%, + 85.10%, + 58.31%, 8 = + 15.46%, + 87.86%, + 62.70%


If you compare the data sets, since the first number in each triplet (Pentium - i3) is bigger than the last number (i3 to i5), the i3 4130 is closer in power to the i5 than it is to the Pentium for gaming, and the other way round for computation.

 

[DonkeyOatie]

If you have more detailed questions on the benchmarks with 1333Mhz or 2400Mhz memory, ask. The 4130 will run 2400Mhz memory on the Z97 board.

 

[joex444]

This isn't great to read, but it's a little bit better. I truncated your numbers because there's absolutely no way you can justify 4 significant digits short of running a few hundred iterations of each benchmark.

i3 vs G3258 | i5 vs G328 | i5 vs i3
GP use
PCMark 7 = +9.5%, + 18%, + 7.5%
PCMark 8 C = + 6.8%, + 26%, + 18%
PCMark 8 A = + 8.3%, + 27%, + 17%
Gaming
IS = +37%, + 76%, + 28%
CG = + 45%, + 90%, + 31%
SD = + 35%, + 91%, + 41%
FS = + 67%, + 105%, + 22%
UH Low = + 50%, + 63%, + 8.5%
Ultra = + 77%, + 100%, + 14%
UV Low = + 39%, + 77%, + 27%
Ultra = + 60%, + 94%, + 21%.
Numeric
1 = - 0.7%, + 18%, + 19%,
2 = - 8.1%, + 19%, + 30%,
3 = + 4.5%, +61%, + 54%,
4 = + 15%, + 84%, + 60%,
5 = + 8.3%, + 64%, + 51%,
6 = + 12%, + 70%, + 52%,
7 = + 17%, + 85%, + 58%,
8 = + 15%, + 88%, + 63%

As a sanity check, labelling the columns as A, B, and C, we should find that C = (1+B)/(1+A) - 1. This appears to be internally consistent.

 

[DonkeyOatie]

This isn't great to read, but it's a little bit better. I truncated your numbers because there's absolutely no way you can justify 4 significant digits short of running a few hundred iterations of each benchmark.

i3 vs G3258 | i5 vs G328 | i5 vs i3
GP use
PCMark 7 = +9.5%, + 18%, + 7.5%
PCMark 8 C = + 6.8%, + 26%, + 18%
PCMark 8 A = + 8.3%, + 27%, + 17%
Gaming
IS = +37%, + 76%, + 28%
CG = + 45%, + 90%, + 31%
SD = + 35%, + 91%, + 41%
FS = + 67%, + 105%, + 22%
UH Low = + 50%, + 63%, + 8.5%
Ultra = + 77%, + 100%, + 14%
UV Low = + 39%, + 77%, + 27%
Ultra = + 60%, + 94%, + 21%.
Numeric
1 = - 0.7%, + 18%, + 19%,
2 = - 8.1%, + 19%, + 30%,
3 = + 4.5%, +61%, + 54%,
4 = + 15%, + 84%, + 60%,
5 = + 8.3%, + 64%, + 51%,
6 = + 12%, + 70%, + 52%,
7 = + 17%, + 85%, + 58%,
8 = + 15%, + 88%, + 63%

As a sanity check, labelling the columns as A, B, and C, we should find that C = (1+B)/(1+A) - 1. This appears to be internally consistent.

The significant digits depend only on the accuracy and precision of the original numbers. The average of five numbers with four significant digits, still has four digits. It's true that the ones with five significant digits should be truncated, but that's Middle School students for you. All of the raw data was measured to at least four significant digits.

Thanks for the reformatting. The original form was chosen to be consistent with the other data collected,