[Ram bandwidth improvement with 8channel memory?]

Hi, how much does memory bandwith affect performance? I'm in the process of building my new rig and I'm considering moving to a cpu with 8 channel memory support. I'm currently running a i9 14900k with dual rank, 2 channel 64gb 6400Mt/s ddr5, with a pice 4.0x4 NVME and I'm still having lagging and program lockup issues, especially when copying/pasting multiple complex symbols. Watching the system monitor when it's frozen, I will see the memory usage slowly increase, once it stops rising it will unfreeze.  This leads me to believe it's my memory bottlenecking the program.  

 

If I were to make a system with single rank ram in 8 channel operation, I would theoretically x4 my bandwidth and also increase my per channel speed since it's only 1 rank per channel.  If my lagging problem is due to insufficient memory bandwith, this should cut my lag times by 1/4,  300% improvement.  

 

Hopefully someone that knows how the program handles memory management and operations can give some insight into my problem.

 

The cost would be significant to build an 8 or 12 channel system, but it would be worth it, assuming it would actually increase my performance.  The real question is whether or not I will see any benefit. 

 

Does anyone have experience with quad or 8 channel memory? Does it offer any improvement?   Thank you.

[New PC build - workstation vs consumer cpus]

Thanks for your comment but i was asking here because the system requirements are very vague. it doesnt mention workstation cpus at all.  i have more than they require for the high end and still see major slowdowns.  Im hoping to see if anyone had some experience with server cpus vs consumer cpus.

[Grouped symbols moving to random places when duplicate moving groups sometimes.]

Hi, sometimes when i make a group of truss and lighting instruments, different components of the group will move to seemingly random places when i ctrl+move the group to duplicate the groups.  when it does it, for example ill make a group with truss and a few lights on it,  then ctrl-move the group and 1 light will shoot off the group and land 50 ft in a random location. this is rare but it seems it the only way around this is to Ctrl+c Ctrl+v then regular move the group.  Is this a bug or do i have some setting wrong?

[New PC build - workstation vs consumer cpus]

Hi, I've been having significant lags when moving and manipulating complicated groups and models, I'm looking to build a new PC in hopes to speed up these lags.

 

My current specs are 10900k w/ 64gb ddr4 and nvme with RTX 3080.  

 

When picking my processor, I'm considering between the 13900ks and a xeon based system, but I have a few questions on what the specs will do for me in real world scenarios. I know max single core speed is ideal, but has anyone compared xeon vs core series builds? 

 

Is the DDR5 vs DDR4 performance noticable in vectorworks?

 

how important is the L1 and L2 cache for performance?

 

Has anyone had a chance to test the new sapphire rapids xeons with the HBM2e memory? Does this provide any benefit to the program? I'm thinking this might be the solution, I notice during the lags, the ram utilization will change significantly.

 

I'm trying to figure out what my bottleneck is and why I'm experiencing such lags, up to minutes of delay.  For example, if I duplicate multiple large and complicated models, the program will just sit there for seconds to minutes, with no activity in my performance monitor, and the program totally unresponsive.  Even closing a file, I will experience significant lag while the program dumps the ram. Is this a software limitation or something I can solve with hardware? 

 

Thanks for reading and hopefully I can figure out a solution.