Due to a recent thread... and additional insight regarding AMD's Infinity Fabric, I decided to re-test the Ryzen 1800x.
Infinity Fabric's speed is directly linked to RAM speed.
Thus, with fast RAM... the 1800x's low-latency audio performance should improve.
 
I set out on what would become a *quest* to run DDR4/3200MHz RAM (at 3200MHz) with the 1800x.
 
I tested all the highest-end X370 motherboards from Gigabyte, Asus, ASRock, and MSI.For RAM, I used the both 8GB and 16GB sticks from Crucial, Corsair, and GSKILL.
None... repeat... none of the eight motherboards I tested would run more than 16GB RAM at 3200MHz.
 
One motherboard would run 32GB RAM at 2933MHz. 
Only one motherboard would run 16GB RAM at 3200MHz.  
Most wouldn't boot with the RAM above 2667MHz.
 
With the one motherboard that was able to run 16GB RAM at 3200MHz, if you changed nearly ANY setting in the BIOS, it would no longer boot with the RAM at 3200MHz.  That same motherboard had a setting to disable onboard WiFi.  Well... I don't know why they included the setting... as it had zero effect.  WiFi was still operational...  
 
Right now, the X370 BIOS' feel like they've been rushed out the door.
On one motherboard, there are dual LAN ports.  In the BIOS, there's no means of disabling either.
If you're running a Ryzen based machine, the first generation motherboards are essentially beta stage.
There are a plethora of small (basic function) issues to be worked out... as well as larger issues such as broad support for DDR4/3200.
 
 
With the motherboard that almost ran the desired configuration, I was at least able to get some meaningful test figures. 
I've also included figures for the i7-6850k (for comparison).
 
General Benchmarks
1800x at stock speed with RAM at 2933MHz

• Integer Math = 43816
• Single Thread = 2067
• Floating Point = 16117

 
1800x at 4GHz with RAM at 2933MHz

• Integer Math = 47379
• Single Thread = 2132
• Floating Point = 17420

 
6850k at stock speed with RAM at 2400MHz

• Integer Math = 29468
• Single Thread = 2335
• Floating Point = 12212

 
Audio Benchmark
Tested with the Reaper version of DAW Bench... using their multi-band compressor (64Bit version) for load... along with a RME Fireface UFX set to a 48-sample ASIO buffer size/44.1k.  Also tested with a Lynx AES/16e.  At a 64-sample ASIO buffer size (48 not available for the Lynx), performance was virtually identical to the UFX.
 
1800x at stock speed with RAM at 2933MHz

• 528 instances of the multi-band compressor without glitches (CPU load @ 93%)

 
6850k at stock speed with RAM at 2400MHz

• 434 instances of the multi-band compressor without glitches (CPU load @ 99%)

 
Running DAW Bench, you can push the 6850k to 99% load without glitches.
The 1800x is good to about 93% load.
Both measured via Task Manager.
 
With the 6850k, there's no significant performance boost using DDR4/3200.
The X99 motherboard had no trouble booting/running with the DDR4/3200 running at 3200MHz.
 
When running audio projects, there's very little advantage to running the 1800x at 4GHz.
Granted, it's only a slight over-clock.
 
AMD is back in the game with Ryzen...
BUT... and it's a big one, the platform is super flaky at this (still very early) moment in time.
As a professional builder, there's absolutely no way (at this moment in time) I would build an 1800x machine for clients.
There's not a rock-solid motherboard out of any of the higher-end choices.
If you go with a Ryzen build, you will be beta-testing.  
With second generation motherboards, I expect things will be far more solid.

Thanks for the follow up Jim. It will be interesting to see what things look like in 6 months when the MBs are tweaked.

Jim Roseberry

...
Audio Benchmark
Tested with the Reaper version of DAW Bench... using their multi-band compressor (64Bit version) for load... along with a RME Fireface UFX set to a 48-sample ASIO buffer size/44.1k.  Also tested with a Lynx AES/16e.  At a 64-sample ASIO buffer size (48 not available for the Lynx), performance was virtually identical to the UFX.
 
1800x at stock speed with RAM at 2933MHz

• 528 instances of the multi-band compressor without glitches (CPU load @ 93%)

 
6850k at stock speed with RAM at 2400MHz

• 434 instances of the multi-band compressor without glitches (CPU load @ 99%)

...

 
previous test:
 

Jim Roseberry

...
Audio Stress Test
Running DAW Bench (Reaper version of the test with its multi-band compressor for load)... here are the maximum instances that each CPU could run completely glitch-free:

• 6850k - 223 instances of the multi-band compressor (CPU load with transport running was 99%)
• 7700k - 180 instances of the multi-band compressor (CPU load with transport running was 95%)
• 1800x - 170 instances of the multi-band compressor (CPU load with transport running was 93%)

...

 
If these two measurement sets can be compared legitimately, the RAM speedup had an unbelievable impact on low-latency audio performance. Ryzen 1800x by a factor of 3.1, 6850k by a factor of almost 2 (1.95). I didn't see the RAM clock spec on the earlier measurements. I imagine the RAM speedup must be a powerful accelerator for the parallel threads. The general measurements didn't factor up nearly as much.

When running the Reaper version of DAW Bench (using the multi-band compressor for CPU load), both the 64Bit and 32Bit dlls (for the multi-band compressor) are included in the plugin folder.
 
**** If both those dlls are present, Reaper will use the 32Bit version. ****
 
I didn't realize that the first go round... so the first set of results are when running the 32Bit version of the multi-band compressor.
 
 

Thats quite a difference, AMD CPUs have always been sensitive with Ram speed.
 
My 6850k is no longer worthy! :(