I am about to buy new hardware to develop in Unity and I want to buy something strong which save me time in compilation, Bake, game debug running etc.
My assumption as developer is to buy VERY strong hardware and save the time in development so the high price of hardware will return itself in time-wise.
does the following hardware will work better and the editor will really utilize it?
CPU: Intel Core i7 5960X 3.0Ghz 20MB L3 Cache OR Intel Core i7 5930K 3.5Ghz 15MB L3 Cache
Graphic Card: GTX980 OR GTX970 series ?
does strong GPU and high graphic card memory will help and make the build/rendering development process faster, including build for iOS/Android or any other platform?
A beast of a GPU probably won’t be fully used up for a mobile game, but for any level serious development, you should always try to have the best hardware money can buy. If anything, it will give you some good headroom to experiment with potentially graphically demanding systems… Cases where lower-spec hardware would just crash, a high-end setup might be able to hold out.
If you can afford it, I do recommend going for the highest-spec workstation you can find.
CPU wise, now that’s something that makes quite a difference… I’ve recently upgraded from a 2nd-gen Core i7 to a 4750K, which I OC’ed to run at 4.6GHz. My compile times dropped from ~1 minute to ~25 seconds, not even mentioning the boost in loading times switching scenes and such.
A good CPU will be useful in just about every part of the job really. From running the game under profiling (or debugging with conditional breakpoints… man those are slow!), to opening the Unity Editor and Visual Studio, to source control push/pull/merge operations… all of that happens faster, and a fast computer = a happy developer.
It will help, but it depends on your target platform. If you are targeting mobile, a better GPU might not perform better at all.
BTW, I really recommend a SSD for Windows, your dev tools and Unity project. It will decrease loading, baking and building times.
Building requires a lot of read/write on your storage device, it’s as important if not more than the CPU. A couple of years ago, I also tough that it was a CPU only thing until I changed my HDD to a SSD. I dont remember the exact numbers, but I had a C++ project that took around 13 minutes to compile, and with the SSD the job was done in about 2 minutes. Conclusion: a faster storage device will shorten your iteration time when building stuff. In unity, the GI lightmaps seems to rely a lot on your drive too.
There’s no “about” thing in discussions like this. Compilers give exact build time. So I can give it to you.
Something is really wrong with your HDD. Disk just can’t have a 6 times impact for build time. Maybe you use some low-speed 5400 laptop hdd with eco options?
I’ve done my own experiment. I took UE4 source and built it on SSD and HDD. This is exact build time for each of them:
HDD: Total build time: 2404,76 seconds
SSD: Total build time: 2334,94 seconds
These are logs if you want: ue4-hdd-build - Pastebin.com ue4-ssd-build - Pastebin.com
3% build time decrease for such a huge project. In case of smaller projects you won’t even notice a difference.
SSD speeds up projects and document opening, programs execution, writing files and overall OS workflow. But it’s just wrong to assert that it speeds up build time for like 6 times.
PS: And I still think this thread is for “Editor and General Support” section rather than this one.
I was referring to building a Unity project than merely compiling code, there is a lot more I/O when Unity is compressing (& copying) art assets etc than when merely compiling code.
When it comes to compilation, I believe a large amount of fast RAM plays an important role (depending on the size of the code of course) - but I haven’t really done any tests to confirm this.
This doesn’t make any sense at all. Upgrading to SSD surely isn’t going to decrease the build time of C++ binaries that much unless the compiler is doing something very very weird like using the disk as RAM or something. That being said, there are things to consider, for example, AV software weirdness, and perhaps you’re not measuring only the build time, but also other stuff, like, for example, copying a very large lib file or something, still, those numbers are just too much imho.
I agree with what you said, except for the low speed eco HDD part. Of course those HDDs suck, but even if one were compiling all the projects on github I doubt the difference from a SSD would be 11 minutes, let alone that proportion SSD:ecoHDD of 2:13, that makes no sense to me. I mean, if a project takes a month to build on SSD, then I can see how it could take a month + 11 minutes on a ecoHDD.
Even in the case of a “project build”, if building a project is bottle-necked by disk access that much, then the builder code is broken. These IO intensive operations should be broken down and use RAM instead. That’s what it is for after all. Any software that uses the disk that much seriously need a revamp.
