Hacker Read

adtac · 2018-09-15 21:37:29+00:00

I know, I'm just using compression as an argument to show how significant 30%+ is.

azhenley | karma 20145 | avg karma 7.2 · | 2018-09-25 21:37:25+00:00

Wow, I wouldn't have expected that. Do you have a reference for the 30%? I'd like to see what is going on to cause that big of a difference.

mrb | karma 22679 | avg karma 6.53 · | 2024-05-09 04:31:31

Oh, you are right. It looks like it's around 9.5% according to a quick simulation I ran.

shaklee3 | karma 2976 | avg karma 1.62 · | 2019-01-28 19:52:22

We can agree to disagree, but I don't consider 30% to be close. Maybe for applications that aren't time-sensitive, but I rarely deal with those.

stewbrew | karma 1407 | avg karma 1.48 · | 2021-05-16 09:46:39+00:00

Good point but you should probably use the prob that it's 75% or higher.

yjftsjthsd-h | karma 28510 | avg karma 2.78 · | 2017-11-08 11:51:31+00:00

In nearly every subfield of computing, along nearly every metric, most people would kill for a 20% improvement.

A processor 20% faster would dominate the market for years. A new compression algorithm that was 20% smaller would be either copied or used by every archiving system. And yes, a website reduced by 20% is significant.

reply

cassianoleal | karma 4331 | avg karma 2.04 · | 2024-05-20 18:52:28

30/186 == ~0.16.

16% is very different from 30%. Still important, but just a little over half as much.

reply

jbooth | karma 10011 | avg karma 3.14 · | 2011-04-04 15:10:22+00:00

30% kinda provokes sticker shock.

What if you went with max (5%, 0.01/MB)? 30% of 0 is still 0, after all.

reply

buboard | karma 6145 | avg karma 1.29 · | 2019-12-05 20:58:46+00:00

those statements sound like incredible cherrypicking

does 1.11% sound a lot an is it statistically significant?

cut page in half from what to what? from 30 to 15 seconds? or from 2 to 1?

improving one second like how? from 12 to 11 or from 2 to 1? and is 2% really substantial?

reply

Judgmentality | karma 4725 | avg karma 3.74 · | 2020-04-04 06:40:09+00:00

The difference between 0.5% and 30-40% is 2 orders of magnitude.

gonzus | karma 389 | avg karma 2.97 · | 2020-09-18 16:53:47+00:00

This kind of arithmetic is always very unclear and causes all kinds of confusion / miscommunication. This is why I always prefer to be explicit about the absolute numbers (maybe in addition to the relative percentages): "it used 100 mb and now it uses 25 mb (which is a 300% reduction when looking at the final result / which is a 75% reduction based on the initial result)".

anticsapp | karma 124 | avg karma 1.32 · | 2020-06-11 01:51:18

Are you sure it isn't 76% with all that hard cold data you're using?

anomaly_ | karma 86 | avg karma 1.06 · | 2024-01-16 17:00:21

It won't be 30% less. It'll be 10% less.

huseyinkeles | karma 657 | avg karma 4.63 · | 2020-09-18 16:37:16+00:00

Is this the lingo that statisticians/mathematicians use?

I’d just say 75% drop for 100mb to 25mb.

reply

saagarjha | karma 56017 | avg karma 2.29 · | 2023-02-05 07:11:35

Definitely not 20-30%. Maybe 10% on certain workloads.

Estragon | karma 4008 | avg karma 2.55 · | 2010-12-03 12:24:13

No, it's about 0.25% of it by number of cables released, to be precise.

rurban | karma 5612 | avg karma 0.84 · | 2016-12-27 23:12:28+00:00

the performance hit is 20% typically. it is significant.

libertyEQ | karma 117 | avg karma 0.92 · | 2017-11-19 00:15:54+00:00

Just to be clear:

60/300= 20%

80/300= 26%

Not much of a difference, IMO. Make your case if you disagree.

reply

doetoe | karma 548 | avg karma 4.15 · | 2023-02-10 19:26:47

That is actually 55% more (or three other way around, 35% less)

commoner | karma 9473 | avg karma 5.67 · | 2021-05-23 19:38:04+00:00

It's actually 42.9% more. 1 / (1 - 0.3) = about 1.429.