In 2022, it was generally an easy choice to go with the precision 3660. But will it hold onto its crown this year with two new systems to challenge it? We will primarily be comparing the CPU performance since the same graphics cards are supported across the range. The only advantage of the larger 5000 & 7000 thousand series desktops is support for multiple graphics cards, which could increase performance for tools such as GPU rendering In SolidWorks Visualize certain simulations and AI or machine learning tasks. If you have the budget, we don't like to just go on paper specs or generic benchmarks, so we have a head to head here with the precision 3660 versus the precision 5860, both with the top 24 CPU option and both with 64 gigabytes of RAM. will also stack up against a 3000 series workstation dating from 2019. To see just how far things have come, this had an eight core Xeon CPU 64 gigabytes of RAM and a quadro RTX 4000 graphics card. Typical of the best balance between cant and simulation tests at the time. Starting with core CAD workflows we tested in SolidWorks, but the performance requirements of 2D and 3D CAD programs such as draft site and CATIA have similar requirements. No matter the CAD program, the majority of CAD operations need to be built in order one feature or step at a time, and hence most of the time you will only see 1 to 2 cores being utilized, meaning that the high speed CPU's give better performance for a typical task. Opening the complex Nemo submarine dataset fully resolved, the four year old Precision workstation took 2 minutes and 10 seconds. Not bad, but the 5860 can improve on that with a time of one minute and 45 seconds. The precision 3660, however, blows it out of the water with the time of one minute and 13 seconds making the 3660 over 30% faster for typical Qantas and 43% faster than the four year old workstation. These levels of improvement held for other core CAD Tasks such as part rebuild. The reason for this difference is a mix of the precision 3660 newer 13th generation CPU and higher raw speeds. We saw speeds of 5.3 to 5.5 gigahertz in lightly threaded workloads like this, whereas the 5860 rarely got beyond 4.3 to 4.5 gigahertz. It's worth noting that we also had to enable the high performance power plan in Windows to get this performance from the 5860 in the default balance plan. This system took far longer to process the data. performance for the 7960 was similar to the 5860, but we'll talk about that in more detail next. So the 3660 retains our recommendation for CAD How about more multi-threaded workloads like SolidWorks simulation? Now first, some bad news. Unfortunately, despite the precision 7960 being available with up to 56 cores, the evaluation unit that we were sent had a lower specification 16 core CPU than the 24 cores in the 3660 and 5860. This performed the same or worse than the precision 5860. So while we don't expect significant differences to the precision 5860 CAD performance. We will revisit and test particularly for simulation workloads and hopefully with that 56 core CPU for our static simulation, we chose a dataset with lots of contacts which can take advantage of more cores. The four year old system with just eight cores takes 5 minutes and 10 seconds. Our 24 core Precision 5860 now narrowly beats this with a time of 4 minutes and 56 seconds, whereas the precision 3660 takes just 3 minutes and 12 seconds, 38% quicker than the four year old system and 35% quicker than the 5860 for our most heavily multi-threaded test CFD using SolidWorks Flow simulation. The results are similarly stark as the complex fluid flow of a mask with 750,000 cells takes over one hour and 14 minutes on the four year old system. The 24 core CPU of the 5860 takes less than half that time just over 34 minutes. But hold on the precision 3660 betters, that again, taking just 31 minutes and 30 seconds for the same task. Close, but still the considerably cheaper 3660 wins. In our test, the only workload that we found, the 5860 took a small lead in was the few remaining CPU based rendering tools. Ultimately, it's about using the right tool for the right task and budget. If your workload can take advantage of a high number of cores needs more than 128 gigabytes of RAM, or can take advantage of multiple graphics cards such as rendering AI & machine learning tasks, then the precision 5860 7960 could be great choices. Budget permitting, however, considering the top of the range CPU with 56 cores costs nearly £6,000 on its own, these systems are unlikely to be the choice for most. For now, the value of the regular benchmark testing we do with real world data is clear. For the vast majority of CAD and CAA uses, you can actually spend less but get better performance by sticking with the precision. 3660 It offers the ideal blend of the highest single threaded performance for CAD and with 24 cores for multithreaded CAE workloads. and if you're looking for the exact specification, we recommend our optimized Pre-Configured desktop workstation builds on our website where we make it easy with a range of specifications to suit different use cases and budgets.