Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.
Showing results for
Show only
|
Search instead for
Did you mean:
This page has been translated for your convenience with an automatic translation service. This is not an official translation and may contain errors and inaccurate translations. Autodesk does not warrant, either expressly or implied, the accuracy, reliability or completeness of the information translated by the machine translation service and will not be liable for damages or losses caused by the trust placed in the translation service.Translate
Hi all little bit about multicore processor usage. I study a lot about muticore processors and multithread aplications. Nowadays processors has 4 and more cores, but try to study, all processor cores have same L3 cache. So when processor core is going to compute something then needs data from memory. Data must be transferred from RAM into CPU. No problem here. If we have one software thread for one processor core, then this task copy data from RAM into L3 cache of CPU. All still looks good, but when you have big assembly you copy a lot of data from RAM to L3 and with one task you occupied all L3 cache of CPU. There is no more space for other data which can need others cores for computing. User can say ok but we want to use all processor cores and put many task on processor. What will happens, processor will spent much more time in copying data from RAM into a CPU because each core needs another data from RAM and compute time, at all will be worse than in single thread mode. Multithreading can be used only in very specific situations when we can operate on same data with multiple threads. As user I think as many other users but as first try to study computer science and after this write something about hardware usage. I am using inventor every day and I see that inventor is using more than one CPU core. Try to create multiple drawing views from big assembly (but check drawing background updates in settings) and you can see how many cores will inventor use. To GPU usage. GPU is specified compute unit. GPU cannot execute same operations as CPU. GPU is good for SIMD operations but not for SISD. So GPU is good for specific operations. Computer games can use GPU much more because computing of each pixel on screen can be done in parallel but as inventor user can you imagine how to compute assembly constrains when subassembly is not computed yet and you don't know how big is subassemly ? You must wait and when subassembly is computed then you can compute top assembly because you know properties of subassembly. Results of subassembly compute are inputs to top assembly compute. Without input information you cannot compute top assembly. Or simplier example. You have part with one cube created with extrude command. To cube you want to drill hole. So you can say we can do this operations in parallel but can you imagine drill to the nothing ? (because extrude is not computed yet and we want to drill). Some operations is not easy to do in parallel, but some can be computed in parallel. One user write here not nice sentences that Inventor do nothing every year only cash and create new release. I can say that it is not true. I see on my eyes datasets which compute is more faster in new releases as in older. I created one big dataset which compute spent more than 30 minutes but on new inventor it spent only 4 minutes ....... without parallel compute so big improvement. If you want to know more about multicore processors and parallel computing you can read http://www.agner.org/optimize/ article Optimizing software in C++: An optimization guide for Windows, Linux and Mac platforms You will see hardware limits of multicore processing and it efficiency. When you are waiting for Inventor compute you can read this article 🙂 To the future please report all datasets which compute spent a lot of time ( or spend more time than user expected), I know one very good man from inventor team which can solve this problems.
We are using iLogic to its limits i must say, playing with a parametrically designed assemble aroun 5000+ pieces and 1 update takes around 5 minutes. If we have a multicore support over there, we'd be working real fast real smooth
I´m sure that Autodesk is working hard on this issue.
Maybe there are technical reasons, why it´s not so easy to use more CPU´s. But hey ! There are hundreds of intelligent people at Autodesk. They will find enough ideas and ways to make Inventor faster. And if they have to write the hole program new..... It´worth it. Thousands and thousands of users will profit......
I have a problem with the rate of drawings. My PC Is Intel i7 6800K 3,4GHz Overclocking up to 4,2GHZ ; DDR4-3200MHz Kingston, AUS X99-AII, GTX 1070, Samsung SSD 950 PRO 512 GB. My Total Occurences in Active Document is 35000 Open Documents in Session is 5500 drawing. Work with the drawings is very slow. Do you have any solution? What graphics card should I use to by able to work without problems ? I Do not have to wait too long when a work with drawings which have many details.
I have an Intel i7-6900K (All 8 cores running 4.5 Ghz OC) running 3200Mhz Corsair DDR4 with 64 Gb of ram with a Titan X (overclocked) and my OS and Inventor 2018 running off of 2xSamsung 960 Pro NVMe M.2 running in Raid 0 in PCIe slot to be in the airflow pathway to avoid thermal throttling. Inventor and HSM both need to take better advantage of the hardware available. (I also have my Inventor hardware setting set to performance to allow my GPU to make the decisions for it)
I feel that with my hardware, there is no reason Inventor should take this long to load an Assembly when hyperthreading this processor like this (obviously not concerned about power consumption) Why is this still taking long? All Inventor files are on SSD and my read/write speeds are over 300 Mb/s. I am thinking this is the software not utilizing my system to it's fullest potential. Is this true? Will do more testing to check core load and GPU load. I understand certain (probably most) calculations must be done in order accurately, I have one of the fastest GPU's on the market, forget accuracy, do it quickly, show me the assembly and update numbers and specifics in the background if possible. Could this be done? Treat it like a video game to an extent, give me enough to play the game and work on how it looks after.
I will do some more testing and comment again with my results.
A large number of Inventor workflows leverage multiple cores today. Leveraging multiple cores is considered for every applicable new initiative we deliver. Setting this to Future Consideration.
IT IS AN EXCELLENT IDEA. I THINK INVENTOR SHOULD BE REDESIGNED SO THAT IT CAN RUN PROPERLY IN LESS EXPENSIVE MACHINE BECAUSE IT IS OUT OF CAPABILITY TO BUY WORKSTATIONS FOR EVERYONE i.e. STUDENTS.
As a student, a ten thousand dollar i9 core is not an option for me. Multicore computation needs to be supported in every way. An Assembly with <200 parts and <600 constraints is crashing every time I change or move any part. S'little redonkulous.
I'd love to see my new computer use all 16 threads instead of 1, Waiting for inventor to load features for 15 minutes is disturbingly annoying, not to say time consuming and aggravating.
Inventor 2018 Professional (Not responding) - 9% cpu usage - 33 % memory usage
I am not sure this needs to be in the idea station.
This is an optimisation issue. For this to be "implemented" Inventor would need no further performance improvements.
Perhaps it needs to be a priority rather than an idea, as any improvements in the software's performance will benefit us.
Personally I think the performance should be a big priority. With many models I try a few solutions until I get to a working component. I don't necessarily need new features to be able to do this, but it would be helpful to go through these solutions quicker.
This was posted in 2013 and its still only Future Consideration??
I know nothing about how this software is set up, how much of it is legacy based, and how much of it can't adapt without completely starting over. So with that said I approach this comment using a logical and common sense mindset.
To me, if Autodesk wants to stop getting requests for things that are already features, and have been for multiple versions on software products like SolidWorks, NX, ProE etc, y'all should for sure design the program to utilize the computer correctly/more effectively/to the best of it's abilities/etc....at the VERY LEAST!!! Conversely, just be honest and let people know that Inventor is not meant for true production workflows if that is where Inventor wants to stay. SolidWorks used to be pretty annoying too, but it's pretty awesome now so it can be done!
It may fool others, but to me, seeing new bloatware in a new release for a product suite makes me think that continual improvement and innovation for real products like Inventor is held below the metric of "most important." Like I said, if you are still getting people requesting features from other "industry standard" programs, implement those features and abilities if you want to remain competitive, and of course, if it does not encroach on I.P. as I know that can be a gray area with software and features.
I totaly agree with the statement above from Janelson33. Stop creating and putting useless bloatware in your design suites, and start designing the much needed performance upgrades in inventor.
Full use of multicore processor is obviously a big challenge for autodesk and I hope autodesk products are capable to use of multicore processor in next few years.
Full use of multicore processor is a big challenge for autodesk and I hope autodesk products are capable of full use of multicore processor in next few years.
If theoretically it can be done, then this must be a number one priority. Now Inventor is full of features that allows you design any structure you want. But for sure it lacks of performance in massive assemblies.
