Hi All,

Here is a new build:

[url]https://axrt.org/development/owb/i386/OWB_2.0-20241009.zip[/url]

Changes:
- address bar gets updated when navigating to next youtube video
- find text panel can now be closed
- option to open network activity panel is hidden
- option to edit context menus is hidden

As we are getting closer to final release I wanted to give you some updated. From the things I still want to look at are:

- enabling JIT for x86_64
- supporting JIT when running on AxRT
- fix network disconnection problems visible on AxRT
- see if I can improve playback performace a bit

Things that won't be solved in OWB 2.0 are:
- network activity window
- network leds
- context menus
- fav icons
- adblock

Order of releases will be most likely as follows:
1) Update (U2) release for ABiv0 i386 AROS containing compoments needed to run Odyssey 2.0
2) Odyssey 2.0 for i386
3) New release of AxRuntime and ABIv11 linux hosted AROS
4) Odyssey 2.0 for x86_64

Timewise, I'd like to complete all of these release in October. It can overflow to November if there are any major issues found along the way.

Also when you have time, please run this benchmark and share your score:

[url]https://browserbench.org/MotionMark1.3.1/[/url]

Here is mine from 64-bit version, linux hosted:

I did a first test on WMware, see screnshot, calculate that I used a DualCore PC with 4GB RAM!
I will also test on a native PC with the small Netbook Acer Aspire One ZG5 (Native Mode)

For comparison, here is results from Firefox on same hardware. Odyssey is currently ~3 times slower than Firefox in heavy graphics operations.

64bit linux hosted (WSL2 Windows 11)

Whoa, I didn't expect that. So what result does your native browser on that hardware?

result on firefox

With FireFox on the same DualCore PC, I only found that it was twice as fast as WMware (60-fps/160.58) !

On the Small Netbook Acer Aspire One ZG5 512MB RAM, see screenshot !

On my HP Z400 native mode 32 bit

again yahoo mail works fine there are no more browser crashes I can safely download attachments and change the message numbering page

On 2.5 GHz i5 HP EliteBook 8470p native attached. So does running VESA graphics really slow things down that much?

Cheers,
Nigel.

I think it's rather the benchmark failing for some reason on your setup. Did you actually see animations during tests or did you see "still frames"?

I found three other benchmarks that possibly give more predictable results. Please try them:

[url]https://chromium.github.io/octane/[/url]
[url]https://mozilla.github.io/krakenbenchmark.mozilla.org/kraken-1.1/driver.html[/url]
[url]https://browserbench.org/Speedometer2.0/[/url]

First two are pure CPU benchmarks, the third is an use experience benchmark.

Here are my results (64-bit version, linux hosted):

MotionMark on Samsun NC10 Aros native 32 bit Atom

deadwood, I don't know if this is one of the things to do

On www.deepl.com Copy-Paste does not work, not even through (Win+V key, Win+C key), if you use 'Paste' from the OWB menu you get a Guru!

The Traduzuine Window is also reduced

Kraken v1.1 on samsung nc10 aros native atom

What total score did you get? It's was at the top of the page, above the details you pasted.

[quote name=deadwood post=6469]@deadwood - I think it's rather the benchmark failing for some reason on your setup. Did you actually see animations during tests or did you see "still frames"?[/quote]

Yes the animations played, mostly quite quickly too.

On my i5 machine: the Octane benchmarjk locks up OWB at the zlib test., the benchmark for Kraken is attached. The third test is 19.7 runs per minute +/- 0.26

Cheers,
Nigel.

@ntromans

Thanks. ~20 for Speedometer is similar to mine, so your web browing should be okish and not as bad as motionmark would suggest.

For Octane, please give it time - even up to 10 minutes. That's how slow it can compute without JIT.

Test of 64 bit version of linux hosted on three different machines