There are some valuable info on Sandvik website about turn milling https://www.sandvik.coromant.com/en-gb/knowledge/milling/pages/turn-milling.aspx They also have available carbide cutters with ground flat (R216.T4)

Thought I would offer an alternative way to approach your machining process to see what you thought. Dynamic milling that shape verse traditional milling. Hopefully it gives you some ideas on different ways to machine shapes like this. Have a good day. Here is a Link to the file. Here is a Video link showing the toolpath in motion.

The problem is not your toolpath. The problem is your tool. Your tool is not a perfectly ground flat bottom end mill from what I am seeing in the pictures. You must have the tool specifically made with a perfect flat bottom grind when ordered. Almost every endmill has a slight taper inward to allow center cutting. However in your case when doing 5 Axis it creates a problem like your seeing. The only solution using your tool is to take more steps for the finish pass, but leave the roughing like you have it. Take OP2 and leave .01 on the floor. Take OP1 and do a smaller step over and more passes on your multi-pass if your going to use the tool you got. Be careful with how much weight and places your taking stock off that face mill. I would leave a bigger fillet radius in those corners IMHO. Please take the time to define your holders.

Last one. This is a spectacular machine. It is only a 35% discount. Not quite the 40% they usually give, but that is likely due to the quality of the components in this build. 256 GB of RAM (@ 2933 MHz DDR4, and it is even the ECC, not that 'non-ECC' cheap stuff...) Precision T7920 Original: $11,082.00 Outlet price: $7,236.00 1 in stock Certified Refurbished Windows 64 Bit Thunderbolt 3 PCIe card - 2 Type C Ports, 1 DP in Dell KB216 Wired Black Keyboard Optical 2-Button Mouse Tower Chassis CL Adapter 6+2 connector to 2x 6+2 connectors RTX chassis Dell Ultra-Speed Drive Quad PCIe SSD x16 card Thermal Heatsink, Single Processor Air Heatpipe Original Price$11,082.00 Total Savings$3,846.00 Standard ShippingFREE Outlet Price$7,236.00 Tech Specs Intel Xeon Gold 6230 Processor (20 Core, Up to 3.90Ghz, 27.5MB Cache, 125W) Windows 10 Pro for Workstations (4 Cores Plus) 1TB PCIe M.2 NVMe Class 40 Solid State Drive 2TB 3.5inch SATA Hard Drive (7200 RPM) 256GB (4X64GB) 2933MHz DDR4 RDIMM ECC Nvidia Quadro RTX5000 16GB, 4DP Dell Outlet Precision T7920 I think this one, is easily the best "bang for the Mastercam buck", of any of the systems I've posted about. 256 GB of RAM. Nvidia Quadro RTX5000, with 16GB of Video Memory, what?!?!

I'd be very curious to see how that AMD Ryzen Threadripper chip performs "in the real world". My hunch is that Intel will eat the AMD's lunch, for Mastercam Performance. While having "128 Mb" of Cache sounds huge, that is the L3 or "spillover" Cache. L1 and L2 Cache are the primary and secondary blocks of CPU memory. The reality is that this particular AMD chip is optimized for programs that run many simultaneous multi-threaded applications. I don't believe that Mastercam makes use of Multi-Threading in the same way that other programs are designed to make use of the Multi-threading. What do I mean by that? Well, a Stock Model for example, might be setup to process as a "separate thread". But, I don't believe that the Stock Model Calculations themselves, are multi-threaded. In other words, the Stock Model is still calculated by a single process thread, once it is allowed to start calculating (could be dependent on other "input", like a previous toolpath). The 'Stock Model Thread' is not hyper-threaded, so it isn't like that Stock Model Calculation is being spread across your 24 Cores/48 Threads. Technically, you could generate up to 48 individual Toolpaths at at time, but you are still limited by several factors: There is a 'databus' between the Hard Drive (SATA or SSD), and the Chip (processor), which is basically 50 times slower at data transfer, versus the clock cycle of the chip. The main "throttle" in processing data is not the size/speed of the chip. It is the size/speed (flow rate) of the 'pipes' that handle the Data I/O between RAM, CPU, and Storage Memory. AMD has historically had much more latency in data transfer between L1/L2/L3 Cache, versus Intel.

All of you... watch and learn https://twitter.com/engineers_feed/status/1103628364684972032?lang=en