PowerVR Series 7

API definition is a collective process where most if not all IHVs aren't impartial. Yes OpenGL needs some housekeeping, but GL_ES needs less of it and no neither OpenGL equals OpenGL_ES nor has the first much to do with the ULP mobile market. For the record's sake in the older days OGL was in my mind for NVIDIA what Glide was for the late 3dfx.

It's unfair to Khronos or any other organisation thereof to blame them for the shape or condition of any API when fingers should rather point in the direction of all involved IHVs without a single exception.

It wasn't in the least of NVIDIA's interest to have in a GL_ES API compute features up to Tegra5/Logan; things don't really change overnight nor does the universe twist by 180 degrees when they in fact had nothing better to present than for 4 consequitive Tegra generations FP20 PS ALUs.

The truth is that we always can use more advanced features and we always could use the best possible efficiency wherever it comes from and developers could always use better APIs. What exactly is each of the IHVs doing for it exactly is the real question here.
 
its basically a slap in the face to the morons of Khronos group who are too slow for GPU vendors and made a too much crippled ES profile
A general rule, when a highly specialized authority is called 'moron', it's usually a safe bet to assume that there is indeed a moron in the room.
 
Interestingly,NVidia have always been one of the biggest barriers to moving the OpenGL beyond it's roots.

Because sending an army of [strike]twimtbp-infections[/strike] helpers to the studios and have them suggesting the use of their proprietary extensions has proven to be more [strike]monopolistic[/strike] practical?
 
Because sending an army of [strike]twimtbp-infections[/strike] helpers to the studios and have them suggesting the use of their proprietary extensions has proven to be more [strike]monopolistic[/strike] practical?

You're getting better to the day :)
 
Interestingly,NVidia have always been one of the biggest barriers to moving the OpenGL beyond it's roots.
I have no idea of what's going on inside Khronos. If Nvidia is too blame, then boooo to them. But I think the group is showing its limit with the current evolution of the API.
It's finally not surprising that we see all these new APIs emerging...
 
I have no idea of what's going on inside Khronos. If Nvidia is too blame, then boooo to them. But I think the group is showing its limit with the current evolution of the API.
It's finally not surprising that we see all these new APIs emerging...

I would want to believe that all decisions taken are based on the results of a democratic vote between IHVs that decide for the shape of any API variant and that's the exact reason why I'd personally blame partially all involved IHVs mostly for any API related quirks.

Things aren't easy for developers that actually get their hands dirty with code: http://forum.beyond3d.com/showthread.php?t=61388&page=5 and while I myself would had figured that certain things solve a couple of headaches here and there (like the ASTC example on that page...) it's unfortunately not always the case. I think it's a far too complex combination of involved parties, interests, timelines etc. that's it's not easy to find who's at fault for each and every tidbit.

Maybe IHVs need some sort of authority to force a few things into shape if decisions are taking too long or viable agreements cannot be reached; for that of course we'd need complete insight in the Khronos preceedings for API definitions which I'm obviously not aware of.
 
Design by committee is always going to have problems with rapid/paradigm change in comparison to dictated direction (e.g. Khronos vs MS approach).
 
If Ryan got anything wrong, it's overwhelmingly because he couldn't understand my weird Scottish brogue as I walked him through some highlights of the architecture during the briefing.
 
Not sure what Joe meant in that tweet. The peak multiply-add rate is the same, plus you get another FLOP from the SFU co-issue.
 
If Ryan got anything wrong, it's overwhelmingly because he couldn't understand my weird Scottish brogue as I walked him through some highlights of the architecture during the briefing.

Uhmmmm there's optional feature level DX11.0 for 6XT cores or does that also belong to the above? :p
 
Back
Top