CPU A with 4 cores, a total performance of around 10.000 points and a single core performance of 2.500
OR
CPU B with 6 cores, a total performance of around 12.000 points but a single core performance of only 2.000?
I wonder because I’m going to upgrade my 5 year old Phenom II X4 system soon to an Intel 4930K, but I’m not sure if some other CPU with only 4 cores but higher single core performance would better with Cubase 7.5.20 64 Bit…
While waiting Scott (JCschild) to come up and give a definite answer …
Rule of thumb: if you are after extreamly low latency figures, you should seek for higher single core performance. In other cases the total performance is a king.
But … since total performance figures in these cases are so close, I’m not sure if 6-core processor will give you any advantage. Coplex tasks (like DAW) does not scale up in linear manner when number of cores increases: 6 cores of performance X does not give you 6 times X performance.
However, the way Cubase works (thread per track) I think there should be some scaling.
Having said that, since my PC is watercooled, I may even go for overclocking that CPU to 4.2 GHz, but I have to check first if this causes electromigration or some other life expectancy issues - I aim to keep the CPU for 2 - 3 years.
Yes, I’m very interested in what Scott has to say.
I will also ask another question in another thread in the meantime… there is one thing that really bothers me…
its getting to be a balancing act. Cubase does scale well (when all is right)
ideally you would want a single 6 core High GHz processor. easiest advice I can give without confusing the point.
the 4930K is presently the best overall for power and cost.
the 4930K overclocked is near impossible to beat without getting into 10-12 core single processor. and even then its dollar foolish
got the same recently, running @ 4.3GHz per core on an asus board, working fine so far.
the biggest issue is with cubase crashing due to ram issues (like it always has done, regardless of hardware).
(i am mostly working in 32bit still, with jbridge-wrapping kontakts for orchestral stuff sometimes… but i still surpass the critical threshold of the amount of ram a x86 program can allocate just using individual [non-rompler] plugins…)
Sorry to say this, but why on earth would you buy the latest and greatest hardware and then cripple it by running x86(32 bit)? Makes no sense.
If you’re serious about stability/performance then use Cubase x64 version along with ONLY x64 plugins. I can’t think of any major plugins apart from URS that haven’t had x64 version for a number of years now. I’ve been running totally x64 for about 4 years.
it makes sense from a purely practical standpoint- i need to finish off projects to a deadline, and they had been started using a 32bit version of cubase, and 32bit plugins. the increased CPU horsepower lets me do that quicker / with more processing. as soon as i get time off to fine tune a x64 system i can make the switch (which is about a weeks’ worth project in itself). i made the comment however because there is a lot of misconception among daw users who either attribute too much importance to the amount of RAM installed, or to processor speeds vs DAW stablity (whilst they still use x86 software).
I clocked that machine now to 3.8 GHz and it can run 12 (!!) voices of u-he Diva in “divine” quality.
Thats far beyond what I expected even in my wildest dreams. CPU doesn’t get hotter than 66°C on air cooling (with a huge cooler), so I may try pushing the envelope a bit further. 4.0 GHz, possibly?
Stock voltage is what I want. I don’t want to kill that EUR 500,- CPU via electromigration in a few months.
Edit:
Just added two FX channels (all Diva instances sending there) with 2caudio B2… still works without hiccups (using ASIO guard, of course… doesn’t work without).
Have you thought about using a self contained liquid CPU cooler?
Their advantages are:
a) Replaces a big lump of metal that places a lot of stress on the middle of the motherboard with a small puck.
With normal heatsinks, I certainly wouldn’t want to jar the case too much or the heatsink might come loose or crack the motherboard. They are usually only braced at their very bottom with their centre of gravity (CoG) well away from their mounting, especially with tower cases where the CoG isn’t even over the mount points.
b) Takes the heat from the CPU directly to the outside, rather than having one or more fans on the CPU heatsink throwing heat into the case, which then needs another fan to draw it out of the case.
c) Self-contained and zero-maintenance, unlike the daisy-chain piping and reservoir needed for full integrated water-cooling solutions. With the latter, I always worry what would happen if a pipe came loose because of damage or improper fitting during assembly, let alone how one adds new components into an existing piping route!
d) Some, like the Zalman I bought, actually throw some cool air on components near to the rear-mounted fan/hat-exchanger block, so it cools the rear memory sockets and bridge chip on the Asus motherboard.
Some gamers have even bought separate ones for their graphics cards. Certainly, I could just take the fan off my new N750OC video card, and use a CPU cooler, but it runs pretty cool as is.