Adblocking innovation

Lenny_Fox

Level 22
Thread author
Verified
Top Poster
Well-known
Oct 1, 2019
1,120
Update on benchmark ublock_perf_thread_answer.md (with AdblockPlus, uBlockOriging and Ghostery devs particpating and agreeing on the benchmark's setup and validity).

The benchmark is done with easylist filter only (about 60K blocks and 49K hides for references)

1587201565139.png


So on average Ghostery processes 17.5 request in a millisecond (1/1000 of a second), AdblockPlus 9.6 and uBlockOrigin 8.25. With this low additional overhead per request it is hard to believe that Google wants to restrict adblockers to 30K rules (roughly 1/3 of easylist) for performance reasons.
 
F

ForgottenSeer 85179

I am probably overlooking the $all, but I can't find it at How to create your own ad filters | AdGuard Knowledgebase
Here we go: Add $all modifier support to the rules converter · Issue #1506 · AdguardTeam/AdguardBrowserExtension

I don't know why AdGuard doesn't put that in their documentation. Maybe open a issue for that

Update on benchmark ublock_perf_thread_answer.md (with AdblockPlus, uBlockOriging and Ghostery devs particpating and agreeing on the benchmark's setup and validity).

The benchmark is done with easylist filter only (about 60K blocks and 49K hides for references)

View attachment 237601

So on average Ghostery processes 17.5 request in a millisecond (1/1000 of a second), AdblockPlus 9.6 and uBlockOrigin 8.25. With this low additional overhead per request it is hard to believe that Google wants to restrict adblockers to 30K rules (roughly 1/3 of easylist) for performance reasons.
Thanks a lot!
Now i wonder why Brave is so often recommend. Also i doesn't wonder that DuckDuckGo is so bad. That's not their main task
 

SeriousHoax

Level 49
Verified
Top Poster
Well-known
Mar 16, 2019
3,868
Update on benchmark ublock_perf_thread_answer.md (with AdblockPlus, uBlockOriging and Ghostery devs particpating and agreeing on the benchmark's setup and validity).

The benchmark is done with easylist filter only (about 60K blocks and 49K hides for references)

View attachment 237601

So on average Ghostery processes 17.5 request in a millisecond (1/1000 of a second), AdblockPlus 9.6 and uBlockOrigin 8.25. With this low additional overhead per request it is hard to believe that Google wants to restrict adblockers to 30K rules (roughly 1/3 of easylist) for performance reasons.
The benchmark is a year old now. I wonder how things have changed since then!
 

Lenny_Fox

Level 22
Thread author
Verified
Top Poster
Well-known
Oct 1, 2019
1,120
The benchmark is a year old now. I wonder how things have changed since then!
Check the releases. but reading the comments from Gorhil (posting more code optimization is not worth the effort) I don't think he will optimize the code any further, the data (rules) is already compressed, so that only leaves javascript runtime. uBO on Firefox uses webassembly so that could influence the benchmark when other extension don't use webassembly or when they don't mind setting the eval permission in Chrome (to use webassembly, which Gorhill understandebly refused as being a stupid reduction of security) the others might gain on uBO in Chrome. You should ask the devs for details.
 
Last edited:

oldschool

Level 85
Verified
Top Poster
Well-known
Mar 29, 2018
7,698
Now i wonder why Brave is so often recommend.
The benchmark is a year old now. I wonder how things have changed since then!

These tests are interesting tech talk but are meaningless to me in real time re: Brave. It just works for me and its plenty fast, with some good privacy features built-in.

Edit: They continually optimize their adblock filtering - but still having trouble implementing their sync and Brave rewards, both features I don't use.
 

SeriousHoax

Level 49
Verified
Top Poster
Well-known
Mar 16, 2019
3,868
These tests are interesting tech talk but are meaningless to me in real time re: Brave. It just works for me and its plenty fast, with some good privacy features built-in.
Hmm right. Check the values. Those are extremely low values and probably not noticeable at all. Also last year I remember Brave saying they improved their adblocker performance by 69% so the value is probably much lower now.
 

9724anon7537

Level 2
Verified
Jun 12, 2018
65
Hmm right. Check the values. Those are extremely low values and probably not noticeable at all. Also last year I remember Brave saying they improved their adblocker performance by 69% so the value is probably much lower now.


To evaluate performance we used the dataset published with the Ghostery ad-blocker performance study, including 242,945 requests across 500 popular websites. We tested the new ad-blocker against the dataset using different ad-block rule lists, and found that with the biggest list, combined EasyList and EasyPrivacy, it offers a 69x improvement over the average matching time, cutting the average request classification time down to 5.7μs. (Microseconds)



Mind you these were early days/tests for their new rust-based blocker engine, things has been improved since then i.e code optimization.
 

Stopspying

Level 19
Verified
Top Poster
Well-known
Jan 21, 2018
814
With this low additional overhead per request it is hard to believe that Google wants to restrict adblockers to 30K rules (roughly 1/3 of easylist) for performance reasons.

Yes, it is very hard to believe G**gle thinks that anyone with a slight understanding of this way adblockers work would believe this.

These tests are interesting tech talk but are meaningless to me in real time re: Brave. It just works for me and its plenty fast, with some good privacy features built-in.

Edit: They continually optimize their adblock filtering - but still having trouble implementing their sync and Brave rewards, both features I don't use.

Agree, Brave is fast when used this way, without sync and rewards. I've only tried using these two features briefly and they seemed to slow Brave down a bit - that is a judgement made without any serious scientific investigation though.
 

Evjl's Rain

Level 47
Verified
Honorary Member
Top Poster
Content Creator
Malware Hunter
Apr 18, 2016
3,684

Stopspying

Level 19
Verified
Top Poster
Well-known
Jan 21, 2018
814
Top picture my FF browser with uBO

Bottom picture my Brave browser with Adguard Adblocker. Adguard Adblocker is a joke!! And don't trust Brave Shields

View attachment 237741

View attachment 237740
I can't say that I'm very impressed with Adguard Adblocker either. I've tried it previously and sought alternatives quite quickly and I am currently using it on one PC with both Brave and FF browsers. My results are not as poor as your screenshot example, but they are not as good as when I use UBO.
 

Stopspying

Level 19
Verified
Top Poster
Well-known
Jan 21, 2018
814
Ok added. Now that webpage is very clean. But the number of ads blocked is much less than uBO

That makes a significant difference, thanks @Evjl's Rain I'm going to have to find some time to experiment with both Adguard and UBO to make comparisons much more thoroughly. "But the number of ads blocked is much less than uBO" - I don't know how you have both Adguard and UBO set up, if it is as similar as possible this suggests there may be a difference in how they count blocked ads, if its not down to one having a larger list of ads to block.
 

Lenny_Fox

Level 22
Thread author
Verified
Top Poster
Well-known
Oct 1, 2019
1,120
That makes a significant difference, thanks @Evjl's Rain I'm going to have to find some time to experiment with both Adguard and UBO to make comparisons much more thoroughly. "But the number of ads blocked is much less than uBO" - I don't know how you have both Adguard and UBO set up, if it is as similar as possible this suggests there may be a difference in how they count blocked ads, if its not down to one having a larger list of ads to block.
uBO/uMA show elements blocked while AG shows request blocked, so AG numbers will be significantly lower
 

ErzCrz

Level 23
Verified
Top Poster
Well-known
Aug 19, 2019
1,222
Simple workaround based on the blocking third party http and images while allowing stylesheet but blocks a lot of it.

HTTP://*^$third-party,image,~stylesheet

1587679498063.png


Still working out what works best for me but the Medium mode guide in this forum seems to work fine. I've just made that adustmen as a workaround to block all the add images. Most sites I use are only https so I might get away with blocking $all but would probably break to many websites.
 

About us

  • MalwareTips is a community-driven platform providing the latest information and resources on malware and cyber threats. Our team of experienced professionals and passionate volunteers work to keep the internet safe and secure. We provide accurate, up-to-date information and strive to build a strong and supportive community dedicated to cybersecurity.

User Menu

Follow us

Follow us on Facebook or Twitter to know first about the latest cybersecurity incidents and malware threats.

Top