I’m building a new PC, and since ZBrush is currently the only program that routinely stretches my PC’s horsepower, I’m building my new PC with ZBrush performance as my primary consideration.
It seems that Zbrush is very good about utilizing multiple cores. Does that mean it will do better with a i7-5820 than an i7-4790?
Is there anywhere were ZBrush performance is published?
Does anybody have any information on this?
|CPU|Price|Passmark|UserBenchmark Overall|UserBenchmark
|—|---|—|---|
Multi Core|Cinebench R11.5|Cinebench R15 Multi Core|
|4790k|$339.99|11445.00|125.00|672.00|9.67|854.00|
|6700k|$359.99|11928.00|146.00|756.00|9.92|888.00|
|1231 V3|$242.99|9833.00|103.00|555.00|7.93|742.00|
|5820k|$378.99|13413.00|105.00|878.00|11.05|1085.00|
Basically, it comes down to whether ZBrush can get full utilization out of more than 6 cores.
Does anybody know which benchmark is most analagous to ZBrush performance?
I used to have a dual 4 core xeon PC, 8 cores in total. In the BIOS I could turn off one of the two CPUs and that would result in slightly more than half the performance of both cores running, as far as I remember. So ZBrush scales extremely well. If you have ever used a program like Handbrake then you will have a feel for how ZBrush scales.
I would imagine ZBrush could utilise a 12 core Mac Pro as easily as a six core i7-5820K. But if you need a definitive answer you should probably ask in the Questions & Troubleshooting forum or on support.pixologic.com.
Personally, if I was going to build a new PC, I would go for the 5820K. It seems like the best value for money and scaling up to one of Intel’s 8 core processors is money best spent elsewhere, IMO.
Good luck and when you do buy a new CPU/PC, please stop looking at benchmarks for at least 2 years. Benchmarks just make you miserable and distract you from pushing vertices around 
Thank you for your response.
I did end up going with a 4790k build, even though I expect that the 5820 would probably be around 15% faster for pure ZBrush work. If you’re interested, I’ll share my findings.
CPU Cost/Performance
| CPU | Price | passmark | Single core | Multi core | Cinebench R11.5 | Cinebench R15 multi |
|---|---|---|---|---|---|---|
| 4790k | $339.99 | 11445.00 | 125.00 | 672.00 | 9.67 | 854.00 |
| 6700k | $359.99 | 11928.00 | 146.00 | 756.00 | 9.92 | 888.00 |
| 1231 V3 | $242.99 | 9833.00 | 103.00 | 555.00 | 7.93 | 742.00 |
| 5820k | $378.99 | 13413.00 | 105.00 | 878.00 | 11.05 | 1085.00 |
Understandably, the performance roughly correlates with the price. Just to illustrate this more effectively, I then compared the benchmark performance based on a hypothetical “$100 worth of CPU”.
CPU Performance Value per $100
| CPU | $100 worth | passmark | Single core | Multi core | Cinebench R11.5 | Cinebench R15 multi |
|---|---|---|---|---|---|---|
| 4790k | $100.00 | 3366.28 | 36.77 | 197.65 | 2.84 | 251.18 |
| 6700k | $100.00 | 3313.43 | 40.56 | 210.01 | 2.76 | 246.67 |
| 1231 V3 | $100.00 | 4046.67 | 42.39 | 228.40 | 3.26 | 305.36 |
| 5820k | $100.00 | 3539.14 | 27.71 | 231.67 | 2.92 | 286.29 |
When the cards are compared for value of $100 worth of processing, the XEON 1231 jumps to the head of the pack in several categories. However, when building an entire computer, many of the components are identical between builds, and that overhead must be considered in price efficiency.
CPU Cost/Performance: Whole Build
| 4790k | $1,163.67 | 11445.00 | 125.00 | 672.00 | 9.67 | 854.00 |
|---|---|---|---|---|---|---|
| 6700k | $1,272.52 | 11928.00 | 146.00 | 756.00 | 9.92 | 888.00 |
| 1231 V3 | $1,082.77 | 9833.00 | 103.00 | 555.00 | 7.93 | 742.00 |
| 5820k | $1,367.77 | 13413.00 | 105.00 | 878.00 | 11.05 | 1085.00 |
I mocked-up several builds, one for each CPU and compared again. These numbers would be even more similar if I didn’t already own a solid graphics card.
4790k build http://pcpartpicker.com/p/ft3LTW
6700k build http://pcpartpicker.com/p/zxFvxr
1231 V3 http://pcpartpicker.com/p/RTWFkL
5820 build http://pcpartpicker.com/p/VcCHgs
CPU Performance Value for $100: Whole Build
| 4790k | $100.00 | 983.53 | 10.74 | 57.75 | 0.83 | 2.51 |
|---|---|---|---|---|---|---|
| 6700k | $100.00 | 937.35 | 11.47 | 59.41 | 0.78 | 2.47 |
| 1231 V3 | $100.00 | 908.13 | 9.51 | 51.26 | 0.73 | 3.05 |
| 5820k | $100.00 | 980.65 | 7.68 | 64.19 | 0.81 | 2.86 |
When all’s said and done, I ended up going with the 4790k. The 5820k would almost certainly be more cost effective for the bulk of multi-threaded applications, but the 4790k would be drastically more cost effective for all single-threaded processes. Those considerations alone would’ve made the CPUs a virtual tie except for the factor that the 4790k is a drastically more common and more popular chip. This adds additional benefit it having better support and having more of the bugs worked out of the complementary components, mother boards, etc.
It was a close call, and I almost went with the 5820, but in the end, I came to the conclusion that the 4790 was a slightly safer investment (if a little bit slower for some of ZBrush’s most processing intensive operations).
Thanks for your reply though.
I wish there were some sort of site that listed actual ZBrush benchmark scores along with the builds which produced them, or at least some sort of pixologic official statement on which common benchmark were most predictive of ZBrush performance.
Take care