killawatt

I just upgraded by CAE/CAD workstation from a 3900x to 5950x. I took a while to decide what to do as the upcoming 7000 series looks as though it's going to be pretty epic, but the fact am5 will not be backwards compatible with ddr4 was a dealbreaker for me. I have 128gb of RAM for running large simulations and this would cost a small fortune to replace in ddr5 format, plus new mb etc. Unfortunately threadripper was out of my budget too, as the quad channel memory would have been very beneficial. I've seen a 20%+ decrease in simulation times with the 5950x however, so it was definitely worth the money, along with much better thermal performance even with an all-core OC.

Any recommendations for said products? Ideally looking for an adaptor with 3-4 drive capacity. They'll be used for running Ansys simulations so will get hit relatively hard under load, and for long durations i.e., up to 24hrs non-stop. Currently running 5950x, Asus Prime x570 pro.

Would love to see LTT begin to address the environmental impact of technology and feature new developments and creations that are helping to reduce it / resolve these issues. As a major informative resource on bleeding edge technology I think LTT could really help (and probably has a moral/ethical obligation) to educate a wider population on these problems. There's also a just a bunch of really cool things happening to nerd out on! Maybe a series specific to this topic could be cool.

Thanks dude! I might try and grab a couple of used 2687w's then if possible. Hopefully this will tide me over until I can afford a Threadripper system...

Hi I'm looking to upgrade my hp z620 cpu's for CAE/CAD purposes. Currently running 2 x Xeon E5 2620's and desperately need all the extra cores and clock speed I can get!! Can anybody point me in the direction, or enlighten me with a list of compatible CPU's? Please let me know if you require anymore system info thanks Matt

Sounds like a suitably pragmatic plan!

Cloud based options are certainly something I need to look into, however I do tend to work with an air gap because of *ahem* certain software issues... I suppose to only real way to know would be to run some 'real-world' benchmarks with different configurations? Can you see the gap closing to a reasonable level with the implementation of gen5 over the next few years? The only caveat being the endurance of an SSD with intense and prolonged read/write...

Looking for any information on the comparable read/write speed of 2666mhz ECC memory vs the fastest gen4 nvme for use in a CAE/CAD system. I know it's not exactly the same... Currently for running large simulations at a respectable speed all read/write is preferably done 'in-core' i.e. entirely on RAM. This can obviously get very expensive when reaching 128gb+ levels, as well as being impossible on some chipsets. Would it be feasable to supplement a 64/128gb system with a 500gb gen4 nvme ssd without taking too much of a performance hit?

Hi, I'm looking to upgrade my now ancient quadro k5000 and want to know whether dual gpu's i.e. 2 x p4000's would be more price / simulation time efficient than 1 x p5000? Can currently pick up 2 p4000's for around £700 so would save a significant chunk if its worth it! and does anybody have information on whether spending slightly more on the new gen RTX Quadro's would be at all worth it? again they can be found for decent money 2nd hand. Currently running 2 x xeon e5-2620's + 120gb RAM in an HP z620 workstation. I use Ansys for simulation with an HPC license. This is purely to shorten solve times, not for graphics in design work or rendering, I'm not so keen to move to Tesla's due to their passive cooling. thanks!