Total that got cut of there was £3,309. Which to be fair given what it allows me to do now will mean it should pay for itself within a couple of years worst case.
Hey all, thanks for all your replies to my previous post about the beefy machine for test renders, i am delighted to say i have gone ahead and ordered the machine after switching the gpu to a 4080 super, and getting a slighty better power supply.
I have also decided to go ahead and double the RAM to 192GB while they are still builing it. But i am getting concerned about cold boots and memory training.
How often does memory training happen? Is it every cold boot? Every manual reset?
The machine will be crashing alot, its just the nature of pushing them hard, and i dont want to be stuck waiting with that horrible feeling of if it will ever even boot at all, the next time i push the render quality a little too high in 3DSMax.
Would greatly appreciate some feedback on this from someone with experience of machines that have alot of RAM.
It really shouldn't lockup the whole machine or BSOD still, 3DSMax plugins don't run as kernel drivers (I hope?). Although the lockups could eventually lead to BSODs if it corrupts files everytime you have to hard reset. The kernel has the final word unless compromised by a driver, it really should be able to kill 3DSMax and leave you with a usable computer.
I don't know Windows enough, but you should be able to configure it so it can't use quite all of your RAM and CPU so there's always a bit of memory and a CPU core available for Windows to function. That would greatly help not being stuck unable to open task manager. Run 3DSMax as a low priority task as well, so that Windows will prefer giving CPU time to literally any process first and 3DSMax gets whatever is left (which should still be plenty).
I'd still make sure to Google any BSOD codes and investigate their cause. Maybe over time your CPU gets a little toasty, or the CPU vendor got a little too greedy with the turbos and boosts, or just can't sustain the load all that long. I had to underclock mine because I had occasional lockups during long builds and it's been solid since.
It's generally preferable to leave some performance on the table if that avoids full crashes. Each crash is a potential corruption of your project's files, a long interruption in your workflow, and an annoyance.