Understanding cpu core batch processing

Hi, I was wondering if someone could help me find more more about cpu processing in wavlab

I’m in process of building a new pc for the purpose for faster rendering of multiple audio files in wavelab

I’m looking to know which part of a cpu processor renders files faster in wavelab, is it core count or clock speed?

My understanding when rendering using the batch processor, is each audio file is assigned a core / thread and is processed using it and it was the clock speed that determined the render time.

In that case I had eyes set on an Intel i9 9900ks which gets me up to 5ghz

But on some Facebook groups I’m being told to go for an AMD Ryzen? And now I’m not sure which one to go for, for wavelab processing?


“ AMD is always better for rendering as for that you need as many cores are you can have, Intel is good for gaming only due to much Higher boost Clock speeds what does not have much matter when rendering.

so AMD for workstation and Intel for gaming.”


Is this the case for wavelab?

Batch processing in WaveLab is optimized with: “one core - one DSP chain process”.
I am speaking about true cores. virtual cores (hyperthreading) has no interest.
IOW, if you have an intel 8/16 cores, it’s good to run up to 8 batch processes in parallel.

But there is another point to take into consideration: the more processes, the more trafic with your media.
This is why you need a fast SSD. In fact, if you are building a new machine, use NMVe ssds. They are truly faster than standard SSDs. There is a big difference.

AMD Ryzen?

WaveLab works fine with AMD Ryzen. I am running on a Ryzen 3970X with much satisfaction.

In that case I had eyes set on an Intel i9 9900ks which gets me up to 5ghz

If you don’t do batch processing, you would get approx 10% faster than an AMD for a single process.

Thank you, what keeps me steering towards the 9900ks is I’m currently on an i7 3.6ghz and I want to see and increase from my current set up and it’s off putting to see that the base speed in the AMD is like 3.6 where as the higher clock on the i9. Just really looking for an increase in what I have

When you consider a single task, what is important to consider, is the boost (not base) frequency, which goes up eg. to 4.7 GHz for the AMD, and a bis faster for the Intel.
But if you run multiple cores, the base frequency has more importance, and the AMD will generally be better.

Thanks again PG, everyone on certain Facebook groups are saying Ryzen. I’m now considering a 3960 possibly a 3970, I’ve have been spending a considerable amount of time looking at benchmarks but they all appear to be video and rendering related and I don’t see much audio stuff.

Again the dilemma I have here is I’m currently working on an i7 4990 k which has a base of 3.4ghz boosts around 3.8ghz ( new 3960
Build is £2500 spend for clock increase of approx 0.7 but more cores. Really I’m looking speed

I do quite a lot of single wavelab rendering from the wave editor and also montage of multiple clips, between 3 and 10 usually, I’m not sure the thread ripper cpu are going to give me as much processing time saving power, for spending £2500 on a new Ryzen build compared to an £1800 I9900k

I see thread rippers get through a premier pro render as it makes use of the cores but is this the same with audio or would an i9900ks get through audio faster on fewer cores

For Ryzen I will still be restricted due to UAD instance limitations and I’m not sure if any other parts of wavlab (apart from batch processor) that utilise multi core rendering (montage? / single render?) or will do in future? Thanks!

If you use a Ryzen, be aware that there is currently only a single mother board that provides Thunderbolt support (Gigabyte TRX40 DESIGNARE, through a supplied expansion board). Thunderbolt being an intel technology, there are more choices with intel motherboards.

Given your use cases, if you target a fast 8 true cores, that will be a nice setup.
But don’t forget NMVe SSDs.

Thanks PG, I wanted to check again. If I render 1 audio file (all master section vst slots filled) from the wave editor, does wavelab utilise 1 core boost speed? Or does it utilise multi cores?

In your opinion would an i9900ks or AMD tr 3960 render this faster provided all other components where equal and optimal?

And for future, might wavelab change it’s core processing to utilise more cores for a single render if it doesnt do this already?

Thanks PG, I wanted to check again. If I render 1 audio file (all master section vst slots filled) from the wave editor, does wavelab utilise 1 core boost speed? Or does it utilise multi cores?

Several cores are involved (DSP, reading audio, writing audio), but only one really matters, DSP.
If you render to multiple file format in the same time, one additional core is used per file format.

In your opinion would an i9900ks or AMD tr 3960 render this faster provided all other components where equal and optimal?

For a single file render, I think you might not notice a difference larger than 10% in favour to the i9900ks.
But if render 16 files in a batch, expect the Ryzen 3960 to be twice faster.

And for future, might wavelab change it’s core processing to utilise more cores for a single render if it doesnt do this already?

When you render a chain of plugins, each plugin much wait for the output of the preceding one. Therefore multi cores can’t be a benefit (no parallel processing). And generally, one plugin is a bottleneck anyway. Therefore, my answer is no.

For your CPU choice, I would say that if your work is mainly about batch processing, the Ryzen is the right choice. Hence the Intel is more sensible.

You quoted UAD: not that it is not possible to render UADs with many cores, else you will soon get out of UAD DSP resources.

Cheers Phil, I am likely to go with this

AMD Ryzen Threadripper 3960X 3.8GHz 24 Core CPU
Corsair Crystal Series 680X RGB High Airflow Tempered Glass
Asus GeForce GTX 1660 Phoenix OC 6GB GDDR5 VR Ready Graphics Card (I do not game, this would be for video editing / possibly OTT?)
VENGEANCE RGB PRO BLACK 32GB (2X16GB) 3200MHZ AMD RYZEN TUNED DDR4 MEMORY DUAL KIT
Corsair 850W HX850i
Corsair Cooling H150i Pro Rgb 360mm Liquid
SanDisk SDSSDXPM2-500G-G25 Extreme PRO 500 M.2 NVMe
Gigabyte TRX40 DESIGNARE Motherboard

Good. Just, maybe 500 GB of SSD is not so much.

Thinking of getting x2 500gb

Something I am worried about is that I use an old RME Hammerfall card,

https://archiv.rme-audio.de/en/products/hdsp_pci_interface.php

I don’t think it will fit in the motherboard?

I don’t think so.
Moreover, on a related topic, I strongly recommend against using FireWire on Windows. For about a year (from my experience), it’s common to see random blue screen. I have experimented that on 2 different computers (using RME FireWire 400). I think something has changed on the Windows side.
Anyway, FireWire appears to come to an end in the industry.

@chris995: Looks like a nice setup! I’m keeping this topic for when time comes for me to upgrade…

On Firewire and Windows: I’m on current Windows 10 and have no choice but to use FW with my ‘interface’/digital mixer (see signature). No blue screens so far, but I recommend to check after every Windows update that your FW driver wasn’t changed. Windows tends to do that, and my setup is most stable with the Legacy 1334 driver. It might work for others too…

This is an interesting post. I’m curious though, PG, when you say it is better to use NMVe HDs, are you saying that the NMVe drive is a better drive to store your programs on vs. SSD, and use your SSDs for data storage - or are you suggesting running nothing but NMVe drives if you have the option?

I have a Gigabyte Z390 Designare board with an Intel i7 processor and it has two NMVe slots. Also, I will add that I had one of my Samsung NMVe drives (250G) fail to write after 5 months owing to a mystery that MS and Samsung could not explain, causing Samsung to completely wipe the drive to ‘fix’ it. The drive was returned, and continues to work correctly, but I opted to load only programs (65G total) on a new 500G NMVe drive in case it becomes ‘overwrite protected’ again. The logic here is that for some reason the 250G (that I had partitioned to 125G X2, BTW) drive hit an overwrite brickwall that the UNPARTIONED 500G unit will avoid (according to some ppl I tend to trust). Eh, FWIW, it was explained to me that the NMVe (and SSD drives, too) as a rule write forward on their storage areas and eventually, when they reach the end, they begin writing at the beginning - and the partition may have put the NMVe into a downward spiral.

Anyway, this was an eye opener for me and it seems, PG, you are suggesting Terabyte NMVe drives? And these will surely have to be partitioned. At this point I seriously question running programs on a partitioned NMVe, and also question using a partitioned one that would store audio data. But I trust your input, too, PG. Are you saying - if the DAW would accommodate it, that you would run partitioned NMVe Terabyte drives for audio work? Thanks.

I am not hardware geek. This being said, I know that a NMVe drive is a “normal” SSD, from the storage point of view, but the access to the data is much faster, because of the direct PCI connection. And I could indeed measure that the speed in several times faster. But of course, you will find tons of similar references on the internet. The speed is especially faster when reading and writing large files. Therefore, video and audio users benefit from that.
One all new Macs, you only get NMVe, this is not by chance.