Switching to Windows.. Help with choosing hardware?

Hi Folks,

i’m thinking about replacing my 2011 15" MacBook Pro (2,3 GHz, SSD) with a Windows machine.
Since i’ve not been in Windows world for a while, i would love to get some help from you guys with choosing the right CPU, especially. Also, i would like to hear any thoughts from people working on Windows 10 in general.

What i need:

  1. Very good VST performance – i would love to get rid of the need to freeze tracks (mainly producing electronic music with CPU heavy VSTs like Serum, Omnisphere 2, Decapitator…).
  2. Low latency MIDI input (being able to work at low buffer sizes) even in soft synth heavy projects.
  3. Should be rather quiet (My MBP is getting too loud when i have to record vocals in the same room, occasionally)

I have distilled 2 options:
a) Razer - The New Razer Blade 2016 (laptop)
Processor: Intel Core i7-6700HQ - 4-Core, 2.6 GHz/3.5 GHz (Base/Turbo)

b) Scan.co.uk - 3XS FWX99 PowerDAW (desktop)
Processor: Intel Core i7 6800K - 6-Core, Overclocked to 4.2GHz

I’m having a hard time evaluating the practical difference in performance boost, i would would get from those machines (eg. by a factor of the 2011 MBP’s performance). I think it has not solely to do with the Clock Speed, but also with the processor generation and kind?

Any help highly appreciated!

I would go for the desktop if mobility is not so important.

This specific one is very promising however 250Gb maybe are a bit small, especially if you are planning to install Omnisphere etc. I suppose you may go at least 500Gb and even gor for an M.2/NVMe Drive if you can afford it.

About CPU selection here’s a chart:
http://www.scanproaudio.info/2016/06/28/intel-broadwell-e-the-new-audio-system-cpu-of-choice/
But again is up to you and your budget. Please note the laptop CPU is not listed/tested. 6800K may be a good choice.

Regarding noise levels, again with the desktop you’ll be more flexible in picking less noisy parts etc. You can pick a low noise CPU fan, maybe a fanless graphics card… etc

Now regarding the difference in performance boost it’s very difficult to evaluate.
Remember you are going from an old iOS laptop to a new Win desktop.
More likely there’s going to be an improvement, but difficult to measure until you go for it.

I wish I knew more when I was building my rig :unamused: :wink:
https://www.steinberg.net/forums/viewtopic.php?f=226&t=89134

I’ve built my own systems since my first PC many years back, and I try to keep my ear to the ground on the latest developments… Whether you go for desktop or laptop should be down to whether you need portability really, but generally I’d expect the desktop to be more powerful because laptops often use cut-down components to enable all that tech to be squeezed into the case. Also they will suffer from overheating more when pushed hard which can reduce their life expectancy and reliability.

Apart from that, I can only really comment on the desktop solution (because of my experience) and I’d say that it looks like a fine solution. As mentioned above, could be useful to move to a 500GB SSD system disk what with all the content these days (although you can move this somewhere else!!). In my system I went for top speed memory, i.e. 3200MHz, but what difference 2400MHz makes I’m not too sure. I also bought a giant Noctua HD15 heatsink so I could overclock quietly - again, I’m not sure how their heatsink fairs so who knows which is better.

Another thing is the graphics outputs. How many monitors are you intending to use? I see your selected card gives VGA and DVI which may be OK for you, but perhaps you’d rather have 2 x DVI or maybe even DisplayPorts? Something to consider, especially for the future. I personally bought a mid-priced 2GB fanless graphics card which worked fine, but later a friend of mine donated an old powerful GeForce8800 card with 2 fans. It is definitely nippier on the graphics side and the fans actually don’t make any noise because the card isn’t pushed so they idle nearly silently.

In terms of latency, that is a tricky point… My system is stable and I can monitor and record at 3ms. I’m using a MOTU 2048 on a PCIe card which is pretty old technology but works very well. I would have thought that you’d get the same with any decent audio box (say of £500 or more). However, compatibility of the PC hardware seems to get in the way for people on this forum. You’d have thought buying a DAW from Scan would remove this potential problem though! You could always ask them for compatibility info if you’ve got a specific unit in mind.

The other thing I did when I got my system was to buy the most expensive (at the time) 8 core unlocked processor. There’s a lot of dicussion about how well Cubase uses the CPU power available and also further discussion on whether it’s better to get more cores or a faster processor. All I can say is that my 8 core is overclocked to 4.5GHz and it blazes along very hapilly. Compared to my previous 6 core PC which I used to max out to 90% even with rendered VSTi’s, I haven’t really gone about 40-50% on this for full mixes of the same complexity, so I’m happy to have spent the extra cash on the big processor. Personal preference of course.

Enough ramblings… I like the look of that Scan desktop. But no experience of laptops other than cheap ones (which I find run simple Cubase projects easily) and macbooks (which I mainly run logic/mainstage on), but I’ll always prefer a desktop for my studio work.

Mike.

Thanks for the useful comments, everybody!

Regarding system components:
The scan link seems to always point to the basic configuration.
I was planning to get a Samsung 850 Pro 1 TB SSD and the Asus X-99 Deluxe II mainboard with Thunderbolt 3 expansion card.

CPU: After investigating the updated chart (thanks, GeorgeV), the i7-6900K seems to be very solid. Only the i7-6950X is rated more performant, but costs around 500€ more.

Graphics: I would like to continue using my Apple Thunderbolt Display and i have a secondary Full HD monitor. I still don’t really understand what the Mainboard’s onboard GFX is capable of and whether or not i can use the Thunderbolt Display at the Extension card’s TB 3 port… Maybe i’ll just get a system without dedicated graphics and see if i need it? Or is it always advisable to go for dedicated GFX card? If so, any idea if the graphics card’s output will be available at the TB 3 expansion card or only if plugged into the GFX card directly (the TB display needs TB and won’t work at a regular display port afaik)?

A more general question on ordering full systems vs diy components:
How difficult is mounting the CPU to the mainboard and overclocking it these days (remaining components will be as simple as mounting few screws and connecting cables, right?)?
I’ve never built a complete PC myself, but i replaced the factory HDD and optical drives in my MacBook Pro with SSDs, which took no more than an hour with the help of step-by-step instructions and i also replaced components in tower PCs and Mac Pros back in the days :wink:.

Furthermore,i’m not sure if there might be difficulties with warranty, because i’m not in the UK (Germany)… Also, i don’t know if the testing they do is a real advantage for me or rather quality assurance for themselves, which they “sell” as a service to the customer (not saying that’s a bad thing)… What do you think? If i have a choice of solid components and the ability to use google, should i still go for the ready built sytem?

@Mike: What’s your processor model? Would like to look up it’s reference benchmark in the chart.

Ps: My Soundcard is a Babyface, so i’m hoping that there won’t be many issues.

Best advice I can give is go for a tested system made by specialists to minimise risk. Plus they will answer all your questions. Aren’t there any daw system specialists in Germany? Would you consider a hackintosh maybe?

I know one DAW specialist da-x.de but they offer very few and specific configurations, which don’t exactly meet my needs. Then there’s caseking,de who are specialised on modding, overclocking and gaming, but also claim to do audio PCs on request and have the individual components i want in their sortiment, so they would be my best bet.

A Hackintosh could be very interesting, since i’m generally very happy with OS X and the mac specific software… Would you mind telling me some more about hackintoshs in general? What to be aware of…, traps?

Thank you!

There is something BIG around the corner, it’s hitting the shops now. It’s called the Samsung 960, it comes in two versions the Pro and the Evo- the Evo being 25% cheaper. It’s an M2 SSD but with a big difference - its five times faster than any other SSD. Way faster than your 950s

An HHD is around 100-250 mbps, an SSD is around 500mbps, this baby goes up to 3.5 Gbs pr second! That’s around 7 times faster than most SSDs.

It comes in 3 sizes, 512 gb, Itb and 2tb.

It will serve samples from a sample drive at an amazingly fast rate, or/and it will make your OS and Cubase blisteringly fast.
The fastest consumer drive on the planet is hitting the shops now. It’s not cheap, the pro version is about £500 UK for the 1TB drive. The Evo is 25% cheaper on the whole, its almost as fast (the second fastest on the planet) but it does have a caveate, the drive slows down after its cache is full- the cache is big mind you, at 44gb on a 1 tb evo, but if you plan huge templates I would avoid
You need a decent x99 mobo and if the mobo has no slot, a card holder. Most X99s have a PCIe slot that it can utilise, it needs PCIe to reach those speeds

The problem with Hackintoshes comes when it’s time to upgrade the OS. Unless you hire someone to do it, you have to research, do text-based config (in files), and spend time troubleshooting. I ran one for a couple years, over two OSes. I think 10.8 and 10.9, but it was not worth my time to do the work of updating. It was fun while it lasted though.

A few more bits:
You can buy PC graphics cards with no fan - silent. If you have SSDs too that means only one fan (the case fan) is required - get a quiet one if you record live.

The X99 Asus mobos are high quality but their support is poor. If you raise a query you have to fill in a form that is very tedious and off topic usually, with details like serial numbers, all version numbers. They take a while to get back to you too and usually with wrong info, because they don’t read what you write well. There is no telephone support at all for mobos. There software for updating crashes on my machines across two mobos and various windows versions. Updating manually is possible, but it involves downloading a cap file, renaming it, putting it on a clean USB stick, then going into Bios and loosing all your bios settings with the update. I may get a gigabyte board next time.

I have one of the first X99s from them and it has usb 3.1 which is apparently twice as fast as USB 3. Make sure you get that cos early boards did not support this.
Although the manual states that the board only supports up to 64gb of RAM (DDR4) after a bios update these boards actually support 128 gigs of RAM. This might seem overkill, but if you work large projects with lots of VSTs you will be surprised.
YOu dont have to start this way of course, you could load less RAM, but if you want to leave expansion room you will need to use 16gb sticks. Asus boards do not support 32 gb sticks- if your wondering, few boards do.

My processor is the 5960X, the purple one on that chart is the closest, but I overclock to 4.4GHz so it would be a bit faster (I did have it at 4.5 but at that speed it didn’t pass the PRIME95 tests). Mobo is an Asus X99-A with DDR4 memory (Corsair 3200 CAS 15 type). I’ve had a few generations of Asus mobos in the past with no problems, they are my favourites. I did have problems with my last generation Gigabyte mobo so I went back to Asus for this latest build.

Actually, I just benchmarked my CPU using CPU-Z and versus the 6950X mine is just slightly better by about 5% (note I’m overclocking at 4.4GHz whereas the reference 6950X is at stock 3GHz).

Re: mounting the CPU in the mobo, that’s a bit daunting for sure given the cost of the processor, and you have to have a steady hand, but it’s not really tricky as long as you follow the procedure, take it slowly and carefully and understand what you’re doing, probably a 30 second job actually. Mounting the heatsink can be a pain too, but again it’s just a matter of being careful and slow as you do it, probably a 5-10 minute job. Then you’ve got to mount the mobo in the case and route all the PSU cables.

Ah, one thing to check if you’re going DIY is that your choice of heatsink doesn’t get in the way of your choice of memory boards. E.g. a lot of large heatsinks will overhang the memory slots and if you’ve just bought memory with big heatsink fins then you may find they won’t fit under the heatsink. And, also, for the same reason you may need to insert the memory before putting on the heatsink… It’s difficult to check this though because you can’t really get the exact dimensions, so I just make sure I buy mobos with memory slots that aren’t all right next to the processory slot, and then buy say 2 memory boards instead of 4. Of course the same goes for the height of the heatsink, i.e. will it fit in your PC case… (I mention this because the bigger the heatsink the faster you can overclock the processor but they do get pretty tall!!).

Mike.

Hi,

Even for Windows, I would recommend an Apple HW, so MacBook Pro.

Modern Core 17’s are very easy to fit, The bios is also much more sohpisticated but has an ‘eazy’ mode, where the machine basically over clocks itself. With modern BIOS there is also a mouse. The fan can be more trouble, but its not realyly hard, just make sure it’s suitable for your mobo

Guess i will let it get assembled for convenience and warranty. Will be my first PC in a while. I can become comfortable with the stuff again and opt for DIY next time.

What’s the best option for Windows 10 (Home vs. Pro)? Any downsides to choosing the cheaper OEM version, since i will probably have that option when i get an assembled machine?

Regarding CPU:
I did some more reading*, and there are several sources indicating, that clock speed is preferable over number of cores in realtime audio processing applications (one channel, including it’s signal flow with Return FX and Group Channels can only be handled by a single thread/core).

Now, i’m confused trying to translate this to the actual CPU model choice :unamused: :
Will a 6700k overclocked to 4.4 GHz actually outperform a 6950X overclocked to 4.0 GHz in single core comparison? If they were both overclocked to the same speed (eg. 4.0 GHz), would they show (nearly) identical single core/thread performance, or is the newer generation’s 4.0 GHz somehow “more performant” than an older generation’s 4.0 GHz?

Additionally, would you recommend a UAD card/interface for the main reason being off-loading realtime processing from the CPU in order to improve overall realtime performance?

For that you will need a decent CPU cooler - luquid cooling maybe.
in other words more fans and of course more noise.

I recommend you post about what you plan to do - musically. If your just recording your four piece in your studio, well anything (practically) can do this, but if, like me, your composing using lots of heavy duty VSTs all loaded at once then you need a lot more, particularly RAM. Top orchestral composers in Hollywood and beyond, often have slaves, (using Vienna Ensemble Pro to link) but things are getting faster and more powerful all the time. I hope to stay on one machine.
The disabling tracks function works very well. I have a whole orchestra (EWQLSO, Halion orchestra & All Spectrasonics, BFD3 and a lot more here, organised into one big template. Its all disabled until I want it so it uses very little CPU or RAM. Most music uses few instruments and if you want a 70 piece string section then that’s usually just one sample set, not 70 different tracks.
There are literally hundreds of tracks ready to go, but disabled and invisible, if I want a scottish bagpipe, its there ready to go, but when the template loads, there is nothing to see except a couple of tracks - an Audio track with a test for my Surround system, and a piano track to check my MIDI. The rest is hidden and taking virtually no CPU or RAM until I click a couple of buttons to activate tracks.
So, what I am saying is you might not need a big machine, there is a lot you can do to reduce loads (freezing, purging, reducing the size of headers) and a lot more besides.
The trouble usually comes when the user doesn’t know how to do this

Still, get the best you can, spend some time learning how to set up specifically for music

Z

I worry about heavy overclocking on a DAW… moderate is fine, but too severe and you can have issues. Maybe this is part of the problem too.

Correct.

I’m producing electronic music of different kinds.

  • Track count: always < 100, probably between 30 and 60 most of the time.


  • VSTi samplers: Mainly simple one-shot sampling (Vengeance Phalanx) for my drums. Additionally, BFD3 in some projects and Kontakt instruments from time to time (but more often going for soft-synths). (I think i’m fine with 16 GB of RAM)


  • VSTi synths: Using Serum and Omnisphere 2 a lot. Massive, Diva, Sylenth and a few others as well, but less frequent.


  • VST FX: Equilibrium and/or Neutron on most tracks for corrective stuff. Some tracks with heavy sound design processing (Soundtoys 5, Amps, Filters, creative FX, LX480 Reverb…). Group tracks for side chaining with LFO Tool (so that i can group tracks utilizing the same sidechain vs having to insert LFO Tool on each track.


  • Audio Recording: Almost exclusively limited to single-track vocal recordings.


  • MIDI Recording: Would like to do it a lot more (esp. with drums), but my current setup forces me into larger buffer sizes rather soon :neutral_face: That’s one of many reasons for upgrading.

I do freeze tracks all the time, but i would like to avoid it when possible. What do you mean by “purging” and “reducing the size of the headers”?

Re Overclocking: I took the oc options from scan.co.uk as a reference, they seem to know what they’re doing.

Let’s say you load a typical, fairly decent sampled piano into Cubase. What typically happens is all the samples get loaded. For each note, there are layers, a sample for ppp, another for pp, one for p, mf, forte, mf, ff, and fff. Sometimes an instrument can have ten or more samples per note, then you might have pedal up samples, pedal down samples, and so forth. As the sampler does not know what samples your going to actually play, it loads them all, just in case. Correction, what Kontakt 5 Halion and many other decent samplers do is load in a “header” - this is just the first part of the sample loaded into RAM, ready to rock and roll, whilst the machine fetches the other bit, when a note is pressed. This gives the machine thinking time and reduces the size of the sample pool for the instrument - in RAM.
After you finished your performance/recording you really need very little of this to remain in RAM. You might have used only a few notes, and only some mf, samples, some pp and so forth - all the rest can get out of RAM until you need them again, trouble is the sampler never knows when you stop recording your masterpiece. This is why a purge button is provided. There are two types, global purge, for all instruments and local purge. Mysteriously, in Kotnakt 5 the global purge has gone missing, this might mean its automated.
Anyway purging is getting rid of header samples no longer required in memory. Moral if the product has a purge button use it, particularly on templates. It gets hard to be vig
ilant all the time about this, but its a good place to look when your getting pops and crackles.
I am no expert on overclocking, I just go from my experience. I think that prolonged running at overclocked rates can reduce the life of the CPU (it gets hotter) under some conditions, Tom’s Hardware or the Geeks on the Hardware forum would know more.

Z