Basically because the modern GPU pipeline is designed around only modifying data using shaders. Shaders have their limitations, so Blender, which was designed over 20 years ago, doesn’t use them to modify geometry. It has to modify the model on the CPU in system RAM, recalculate any and all modifiers, and upload it to the GPU every frame when something changes.
Note I’m not saying Blender’s edit mode can’t be made faster, just that DCC apps in general are a poor fit for the modern GPU pipeline, so naive implementations will tend to perform poorly. Speeding it all up will require some smarts and trickery and will probably make maintenance harder, so I’m not surprised they’re putting it off until they have a better picture of what the final feature set is, and where the bottlenecks are.
Let’s hope they will boost edit mode because I know some companies who don’t use Blender because of his poor perfs.
And it seems 2.8 is worst for now than 2.79 who was already a no go.
so i run some tests on my machine using the 2.8 demos, the workbench engine, runs really super fast in object mode like 6-10ms which is equivalent to 100-160 FPS even on mesh with 10.5 million tris but Edit mode slows down really hard and EEVEE also good but could be better the frame drops just little bit, i hope there is room for more optimization and improvements.
I still can confirm that 2.8 blender is not production ready in terms of high fidelity production. For example if I would to edit high poly mesh to fix some errors I could’t because blender would slow down to a cramp, or crash. Or worse crash the computer.
Yup. 2.8 won’t be production ready for a long time and it’s entirely unfair to expect otherwise. Don’t let the forum’s enthusiasm affect you. It’s a bug-riddled, slow, half-baked mess at the moment.
For example making AAA quality meshes. You need high and low poly in your modeling software and be able to manipulate it. If you can’t do that blender 2.8 won’t go anywhere near that kind of usage. And I check blender if it’s ready for anything of sorts when someone mentions blender got performance update. I know blender is in Alpha 2 stage now, but when performance optimizations come into play? No one is talking about that and that is most important part for modeling.
stupid me, I realized this tread only now … the question of performance and optimization is something that is very close to my heart …
I believe that eevee has much room for optimization, in some scenes switch off or turn the shadows on my gpu performance halves two or three times … it is clear that it is an optimization problem
especially I care that when I move an object, give priority to the object, even turning off for a few thousandths of a second shadows and lights if necessary, as long as the object is moving.
Once we hit beta and have the feature set dialed in, then there will be more of a focus on optimization.
The alpha is currently pretty rough, and the beta is also going to need a lot of work done on it. Frankly I wouldn’t consider blender 2.8 for heavy production work until 2.81 or 2.82. Optimization takes a long time when you only have a handful of developers.
We do have a volunteer developer looking for places in editmode drawing to optimize as of now.
As for speed issues in Eevee, Eevee’s graphics as of now are geared towards being high quality as opposed to super fast (which is what makes it capable of such good results). There might be a need for the devs. to add a “fast” setting that degrades visual quality, but vastly speeds up drawing.
Most of the performance issues people are concerned with are in edit mode. The solid shaded workbench mode is pretty fast, until you go into edit mode. Eevee can be a little slower, but it doesn’t need to be super fast.
As it stands right now, there is no fast edit mode.
i get ur point, and i absolutely agree, but i don’t think optimization will come any time soon, there are many things to be finished and integrated into 2.8, this year has been a crazy for the devs from moving to a new office,relocating , Code Quest,Crash fixes, and now Blender Conference / Beta, it’s too much to ask for a small team…but after a locked beta the community will have a chance to test drive 2.8 on different setups ,and with the new open data developers will have a general idea on how to make great optimization that works on most conditions.
I exaggerated the “shadows cube size” and “cascade size”
to make clear another trick where possible to optimize by automatically scaling these two factors during the selected object is edited
I suggest everyone to investigate the possibility of simple tricks of this kind on where they could make improvements in terms of performance in the editing of objects to ensure that the interaction is always realtime. Think and get ideas and anything else to recommend to developers
I don’t know anything about software development so my opinion is based on ignorance but I kind of wish that the blender devs would drop everything else after official 2.8 release and just focus primarily on the fundamentals like viewport performance. They keep adding more features I’m worried that they’re just going to spread themselves too thin into oblivion. If the viewport is not up to par it doesn’t matter what bells and whistles you add to the program. I’m sure they understand far better than I do but this is just my two cents.
