autocad vs gpu

[thelittlesniper]

Sorry for the account problem, it's my brother's old profile that mistakenly remained synced, I'll see to fix it as soon as possible.
resuming the speech, who uses autocad and has a picture video card (or opengl) finds a better use of gpu?

Not really. I have tried various dwg files on both signing machines (quadro and geforce) and there are no important differences, indeed, if you make a geforce better at that point.

 

[gil]

Sorry for the account problem, it's my brother's old profile that mistakenly remained synced, I'll see to fix it as soon as possible.
resuming the speech, who uses autocad and has a picture video card (or opengl) finds a better use of gpu?

My experience on this is very limited. the only concrete evidence made for a reasonable time dates back several years ago. a painting 600 vs gtx 750, on inventor.
from the specifications the gtx is much higher, also because it belongs to the next generation. in practice the painting sucralassava. but I can't explain why. I don't even think it's an exhaustive test, since it's limited to only two cards on one software.

when talking about hardware one thing for me inconceivable, is that a kid who wants to play some 50 euro video games, can take advantage of many tests, made with the most diverse configurations, by many reviewers. when a company has to run software from the cost of thousand euros per year, it has no equivalent sources of information and basically, it buys what passes the market. so blind.

on youtube and similar some (just) tests there is, but they seem unreliable, made by people who in the specific cad, understands us as my grandmother. Just say I've seen a test on a model, in which venic rotated to the madness and count the fps as it was a video game.
some reviewers like blink46 and saddytech in the past have tried the picture on the software premiere, referring to much better results than gtx. results that were not replicated in gaming, where the gtx dominance was remarkable. but they too, said they did not understand the mechanism and since they are not their reference market they abandoned the speech.

Basically I suppose that any claim you make is not based on verified results, but on personal opinions, since very few or perhaps no one could do comprehensive tests on different hardware and software. the exceptions to this rule I think are very few or none.

Does anyone have any concrete evidence?

 

[thelittlesniper]

My experience on this is very limited. the only concrete evidence made for a reasonable time dates back several years ago. a painting 600 vs gtx 750, on inventor.
from the specifications the gtx is much higher, also because it belongs to the next generation. in practice the painting sucralassava. but I can't explain why. I don't even think it's an exhaustive test, since it's limited to only two cards on one software.

when talking about hardware one thing for me inconceivable, is that a kid who wants to play some 50 euro video games, can take advantage of many tests, made with the most diverse configurations, by many reviewers. when a company has to run software from the cost of thousand euros per year, it has no equivalent sources of information and basically, it buys what passes the market. so blind.

on youtube and similar some (just) tests there is, but they seem unreliable, made by people who in the specific cad, understands us as my grandmother. Just say I've seen a test on a model, in which venic rotated to the madness and count the fps as it was a video game.
some reviewers like blink46 and saddytech in the past have tried the picture on the software premiere, referring to much better results than gtx. results that were not replicated in gaming, where the gtx dominance was remarkable. but they too, said they did not understand the mechanism and since they are not their reference market they abandoned the speech.

Basically I suppose that any claim you make is not based on verified results, but on personal opinions, since very few or perhaps no one could do comprehensive tests on different hardware and software. the exceptions to this rule I think are very few or none.

Does anyone have any concrete evidence?

in reality tests and benchmarks there are but not specific for autocad unfortunately. Usually the various tests give a score based on the time used to make a loop or the number of "rendered" fps and autocad does not work like that! from here comes the difficulty in making an ad hoc test.
if you want there are free software that make you various tests with professional programs based on cad the most used is specviewperf or more specific for autocad cadalyst.

I had found comparative tests between video cards, various frames with various geforces, and in autocad there were no differences in productivity if not for minimum percentages.
this - quadro cad - is a link, a bit dated, but where it is noticed that a k6000 from 3000€ and with specifications at the top, is beaten great by a k4200 from 400€ with clearly lower specifications (in autocad 2d environment). and are cards of the same generation tested on the same machine and with the same program!!

 

[lattu]

Thank you. News Now I'm officially confused.
apart from the jokes I think you're right, I had already looked for comparative tests finding many different solutions, when then I decided to configure the pc many of the selected tabs I didn't have at disposal, so I opted for the rtx3060 because it seemed like a great compromise but obviously my inexpression led me to commit several errors.

 

[TAURUS77]

ciao lattu, provo a cercare una soluzione suggerendoti di installare questi driver nvidia studio
versione 511.65 whql
sono della titan rtx ma compatibillissimi con la tua scheda.

nvidia studio drivers provide artists, creators, and 3d developers the best performance and reliability when working with creative applications. to achieve the highest level of reliability, studio drivers undergo extensive testing against multi-app creator workflows and multiple revisions of the top creative applications from adobe to autodesk and beyond.

 

[thelittlesniper]

even if I doubt they will give me a solution I tried to contact autodesk assistance... Let's see what they say.

 

[lattu]

for the moment I have to interrupt because in days I will send the pc in assistance for the problem on the cpu (well more serious).
when returning I try with the suggested drivers

 

[thelittlesniper]

as I said last week, I contacted the autodesk assistance and after a lot of attempts and solutions proposed by the technical support I did not the same a spider from the hole!!
neglecting the fact that it is necessary to repeat 20 times to be understood and continue to give specific software and hardware; they gave me different possibilities from "average user" of the type:
- Is autocad updated?
- use the integrated video card or the dedicated video card?
- Have you tried the latest video drivers? and those autodesk certificates?
Does it have active hardware acceleration?
- did you clean temporary files?
- did you clean autocad cache (cachemaxfiles)?
- set the blockmrulist variable to 0?
Did you try to disconnect a monitor? : :
all things I had already done but that they have not solved anything. then sent me the link for autodesk advice - performance by autodesk - that's a little what's written on it.

finally sent me a link inviare feedback to autodesk I wrote and sent but who knows if and will take into account..... Maybe make yourself feel even if you have slowdowns, blocks or crashes of the program!

so nothing done, autocad is so point and just! we pay and thank you in short.. .

 

[Tristan]

as I said last week, I contacted the autodesk assistance and after a lot of attempts and solutions proposed by the technical support I did not the same a spider from the hole!!

I break a spear in favor of autodesk. I've been in the channel for almost 20 years, I have been helping and training on various products and I assure you that (although the trend is fortunately improving with the new generations) the vast majority of autocad users can not even do a google search. you have no idea the times when they contacted me for a problem and a few seconds later I sent him a link from home | autodesk knowledge network with the solution to the same. and it is also for this reason that many users, who find a problem, do not even know how to report it to those who have to. and if the problem is not reported how can autodesk fix it?
in my career I didn't find many bugs, but someone found it (also worthy of those who reported the problem) and I assure you that if of bugs it is (so replicable every time, and not only on a particular file maybe damaged or corrupt) the resources to fix it there are.

 

[Tristus]

is a 'autocad. problem of software architecture presumed. more than a problem I would say a lack.
Try filling a layer of 3d tiles and you'll see you can't work anymore. you will be forced to transfer the work to 3ds max, which will have no problem in managing all those polygons (unless you exaggerate with the number of bonfires). personal experience.

cpu or gpu is a separate speech and some consideration should be given.
in the modeling phase what you really need is to have a cpu with a good clock speed, regardless of the number of cores of the processor, because the calculation during modeling is performed on a single core. so a cpu with a high clock speed will easily exceed a cpu with a high core number but with lower clock speed.

I tried to sum up a bit, but this link is explained more in detail

Modellazione 3D e Rendering | Insane Build

CPU Rendering Il rendering della CPU utilizza tutti i core al 100%, ciò significa che se utilizzerai la tua workstation solo per immagini e animazioni con

insanebuild.altervista.org

 

[gil]

qui un link a video spiegone on the difference between picture and gtx. Of course, with the cabbage that shows in-depth comparative tests.

 

[Tristus]

autocad pawns, we know them well. Let's consider it one of the best (if not even the best) replacement of the old tecnigraph. unbeatable in the 2d
As for other applications, for what I managed to understand, I would say that gpu only helps when the program asks for its help. but not always does.

in the real modeling phase is the cpu working. According to the link I posted before, the necessary calculations at this stage are performed on only one core. so having a cpu with many cores doesn't mean fluidity anymore. probably the gpu is called into question when there are various views or refresh rotations.

in rendering phase is fundamental hypertrheading. I suppose we should talk about hyperthreading when we see that the cores of our cpu work all at once. at that point having more core makes the difference
in rendering gpu is of fundamental importance if we operate rendering engines like octane.
If instead we opt for a more traditional engine like vray, maxwell render, crown maker, then there is no need to have a monstrous video card.
I at least do not feel this need, and my very dated geforce gt620 from the age of stone does not complain at all.

 

[thelittlesniper]

in the real modeling phase is the cpu working. According to the link I posted before, the necessary calculations at this stage are performed on only one core. so having a cpu with many cores doesn't mean fluidity anymore. probably the gpu is called into question when there are various views or refresh rotations.

but the point is that I have never seen a single core, or the whole 100% cpu with autocad, and that is what in my opinion should optimize to be more productive.
idem for gpu, never seen working more than 20% (and are good).

so much worth, as you say, staying with an old workstation if you have to turn only autocad but professional applications should not reason like this, should instead be the most scalable possible depending on the pc... see render various: I have an old pc, I know that I will employ 24h to render; I have a new pc from 20k €, make it in 1h.

 

[Tristus]

but the point is that I have never seen a single core, nor the whole 100% cpu with autocad

I'm not an expert, but I think you'll see the cpu work at 100% (maybe with all the cores involved in the calculation) when autocad launches the rendering. I think it's the rendering engine that will require cores to get into operation.
normally, in the modeling phase, it is only a core to work and it is not necessary that you work at 100%.

then I can't say how many video resources are needed during modeling with programs such as cat or creo etc etc., for which programs, for example, a good video card could mean more fluidity.
cores come into operation simultaneously when the program asks for a great deal of work. I think about rendering or photogrammetry programs. otherwise during modeling, the amount of work for the cpu is not so high and the cores are quieter.

I also expected, when I bought the first multicore pc, that the cores worked together in all situations. instead it is only in certain situations that intervene simultaneously, and only if the program calls them into question. i.e. in practice it is only if the program calls them into question.
if Hypertreading is provided in the program instructions.
I repeat, I'm not experienced, I don't want to throw myself into things that I know little, but I think I understand how they work this way.

but professional applications should not reason like this, should instead be as scalable as possible

In fact, I was talking about catia, or creo, or solidworks etc etc., compared to autocad have better software architecture.
the fact that with 3ds max I manage to video 4 falde full of 3d tiles... while with autocad I can barely fill one (but then if I don't freeze the tile layer I can't work) says long. Well, I've got an ultra-thin PC, but 3ds max c la fà, autocad no. I think it's not a question of video card.

 

[gil]

Can you make a video (also with the phone) of the monitor while presenting the problem? otherwise it is not possible to fully evaluate it.

ok, autocad is not the maximum of optimization, but it is not even imperfect.
I would like to see what type of models you use and how severe slowing is.

 

[lattu]

the pc was in assistance, now it's back, soon I'll make a video.
during the renders the cpu works very well but reaches high temperatures (I have to replace the heat sink).
@tristus the core you're talking about is cpu or gpu? because in the modeling phase, if I monitor resource management, it's pretty subtone, but I want to do more tests.
I also noticed that the same model 3d studio manages it much better than autocad but I was convinced that a good video card would fold this gap.

 

[gil]

I hope to see your updates soon.

I was convinced that a good video card combed this gap.

up to a certain point you, the card helps. but beyond a certain limit is how to put a last-generation reaction engine on the triple red baron.

 

[Tristus]

the core you're talking about is cpu or gpu? because in the modeling phase, if I monitor resource management, it is quite subtone.

the cpu. practically only the processor.
I installed the speedfan program and noticed that the temperature of the gpu (i.e., we talk about video card) increases when for example you turn the views with autocad. especially the orbit continues. But my card is really poor, a geforce gt620 2g.

Anyway, I'm afraid with autocad, even placing a good video card, you can't get more than that. the fact that, on the same computer, 3ds max manages a decidedly higher number of polygons than autocad suggests that it is the motor of autocad not to be up to.

the comparison made by gil Azzeccatissimo

 

[GiGa]

But something is moving.. .3d graphics (new version 2023)new graphic engine that "would" exploit gpu and cpu multithread. still not used in all areas (for example it is not exploited in paper space), but it would seem the right way.. .
Has anyone tried?

 

[lattu]

It's a bit of a rat, but it's better than nothing.
Maybe in a couple of releases they will start to do seriously