Direct benefits of VTd over VTx, worth the upgrade?

[Kougar]

Please correct me if I'm wrong! My understanding is that direct I/O overhead resulting from Guest OS's will show up as kernel overhead for the host OS.

I can run 12 Windows XP VMs on my 4770K, and at full load the system is pegged at 100% CPU usage, around 80% of that is marked as red kernel overhead in the task manager.

Is there any way to tell how much swapping my 4770K out for a 4770 with VTd support will benefit this workload? Most of the load is from VMware Workstation's vGPU acceleration but I'm sure there's plenty more from 12 guest OSs.

Processor load is my sole limiting factor, and as my 4770K can no longer OC above 4Ghz I have no use for a crippled "K" part that is missing VTd and TSX support. But I don't want to buy an entirely new CPU if the benefit will be minimal, and I don't have the option of testing one to find out. Any input would be welcome, thanks!

 

[Hugo Drax]

Vtd requires bios support and non server or workstation motherboards typically do not support it. Also VMware workstation does not leverage vtd, you need to run VMware esx on the box and run your hosts on top of that.

Swapping the chip will not provide any benefits for what you are trying to do.

 

[Kougar]

Most Z87 boards do have VT-d support in UEFI, or at least I can personally vouch for Gigabyte here.

Good catch, didn't know Workstation 10 still didn't support VT-d. Had thought the option just wasn't presented as it wasn't supported by my processor.

Based on more research.... Hyper-V is free, and will run on top of Win 8 Pro. It has full VT-d support and just like VMware has its own awesome convert-everything to Hyper-V program.... I may go this route just to play around with Hyper-V and Win 8.1 to get a feel for both.