Gigabyte GA-A75-UD4H Socket FM1

[cadaveca]

With AMD's new and affordable Fusion APU processors now on the market, Gigabyte steps up to the starting blocks, ready to win the race. Sporting a ton of features, the GA-A75-UD4H is not only ready for endurance racing, but superfast sprints down the drag strip too. We strap our AMD A8-3850 APU in Gigabyte's GA-A75-UD4H, and take it around the test track.

Show full review

 

[yogurt_21]

just an fyi your math is a bit off on the oc comparisons, the wprime results are actually more significant than the super pi results.

also the super pi gains are just under 20%, not 25%.

http://www.csgnetwork.com/percentchangecalc.html

always rememeber it's original number minus new number over the original number.

in the case of super pi it's (1322.49-1061.94)/(1322.49) which = a 19.697% decrease in value or in this case a near 20% performance improvement

then the wprime oc change would be (440.21-342.44)/(440.21) which = a 22.21% decrease in value or in this case a 22% performance improvment

this of course makes the latter part of the following statement

SuperPi, when overclocked, completed over 5 minutes faster than at stock, a near 25% boost overall.

as well as the first part of this statement

wPrime, on the other hand, didn't show such significant gains, however still did get a sizable boost.

incorrect.



total nit-pick in an otherwise great review, but there it is.

 

[cadaveca]

I fixed it for you, although probably not in the way you think.

 

[ifkopifko]

in the case of super pi it's (1322.49-1061.94)/(1322.49) which = a 19.697% decrease in value or in this case a near 20% performance improvement

Actually, overclocked system takes 19,7% less time to complete the task, but based on those numbers, it is 24,5% faster. That means a cca 25% boost, so that part is correct in the article.

With wPrime, there sure is some mischief in the article.

 

[cadaveca]

With wPrime, there sure is some mischief in the article.

 

[yogurt_21]

Actually, overclocked system takes 19,7% less time to complete the task, but based on those numbers, it is 24,5% faster. That means a cca 25% boost, so that part is correct in the article.

With wPrime, there sure is some mischief in the article.

original value must come first, it's only 25% if you compare the amount changed to the overclocked number. That is an incorrect way of doing percentages.

1322.49 is the old value or the original value

1061.94 is the new value

 

[ifkopifko]

yogurt_21> You fail to understand, that saying:
a) solution A is X% faster
and
b) solution A needs Y% less time
are two different things.

For example, if stock system takes 100 seconds and overclocked system takes 50 seconds, it is correct to say: overclocking provides 100% boost, because the overclocked system is clearly twice as fast as the stock system. But your "correct math" says that the difference is only 50%. Those 50% are only aplicable in this way: OCed system takes 50% less time to finish the task. Both are true.


cadaveca> I only meant that this statement "wPrime, on the other hand, didn't show such significant gains, however still did get a sizable boost." sounds like OCing doesn't provide such a great benefit compared to superPI, but the relative gain is larger. Nothing more.

 

[cadaveca]

I was referring to the time saved for myself personally. 5 minutes off of a benchmark repeated a minimum of three times to reduce natural deviations in testing, means that's saved me a minimum of 15 minutes in waiting.


Pretty impressive gains from overclocking...not the best, performance-wise, for hte platform, but it's pretty damn good considering cost and intended market.

I knew what you meant, and you know, you're kinda right, and hence my post.

 

[happita]

Is the max TDP on that thing come stock at 111w ? Or is that a result of the OC in that CPU_Z shot at 3.64GHz

 

[cadaveca]

Is the max TDP on that thing come stock at 111w ? Or is that a result of the OC in that CPU_Z shot at 3.64GHz

I do beleive that is default.