Please, anyone have Mac Pro 12 cores to make a filter forge benchmark?

I remember that I have seen that some of you have a Mac Pro computer and with dual CPU, like Totte that have one, but I think it was with 8 cores.

Please, anyone have Mac Pro 12 cores to make a filter forge benchmark test?

ALSO would be the same if anyone have a Dual Xeon 6 cores (12 cores) 2 Ghz or 2.2 GHz on Windows as it would be the same

----------------------------------------------------------
----------------------------------------------------------
----------------------------------------------------------

I want just to ask a very simple render of a 3000 x 3000 and 4000 x 4000 render time of some filter that you may like and have and that is not a photo filter, better if it was a FF self generated image but also can be other filter.

I just want to know how much time does it take to render at this resolution on a 12 core computer.

----------------------------------------------------------
----------------------------------------------------------
----------------------------------------------------------

As on the internet there is NO reference in any web about any benchmark made with filter forge, I want to know the difference in perfomance from a dual cpu 6 cores.

I know that the cores alone are not important, and what is really important is the perfomance and real world speed.

I know that the MacPro 12 cores are based on two 2.4Ghz Intel Xeon E5645

.
NOTE
.
I have made a copy and backup of this post it is lost as it happens lately

.

I have FF in it soon too.... Will test with some of my filters, can give you 12 Core plus 8 Core, with default AA settings. I'll be back!

used: http://www.filterforge.com/filters/9225.html

First preset, 4000x4000 11 minutes sharp.

Quote
Totte: Baah! The "Slow elapsed time" seems to be buggy in 3.0. When I render a 4000x4000 to disk, the elapsed time shows the time it took to render the preview after the render was done

This happens only when you have the "reduced" preview activated instead of the "Actual" preview, so whatever size you have put it will always show you the time of the 600x600 render preview.

So to know the real time of the 4000x4000 render you must put the "Actual" size preview.

Skybase, I do not know if you have seen that my computer is ONE CPU Quad core and the one fr om Totte is TWO CPU quad core (or six core) although is very important that they may have different GHz value so it will not be probably double speed.

I mean that I think is not possible that my Quad Core 3.4GHz is FASTER than a his MacPro 8 core computer even if it has a lower GHz value.

WHY I AM ASKING THIS HERE?

I have seen that having more cores DOES NOT mean that it will be a better and much faster computer, and directly computable and multiplying perfomance.

I have seen some cinebench benchmarks wh ere a Quad core (8 core threads) like mine has a 7.5 score AND a Dual Xeon 12 core (24 core threads) has a score of 14.5, it should be triple the score instead BUT it is not because the Dual Xeon have each 2.2 Ghz instead of 3.4 Ghz like the Quad Core.

And about cores, it depends on the task and how well the code handles many cores.

If the code uses a lot of signals to synchronize the threads, about 4 or 5 is a break-even when each added core will only add marginally as it will steal so much computing time by adding more and complex synchronizations.

I've been programming multithreaded and multicore for almost 15 years so I know most of the traps you can step into.

They are 2.4GHz CPU's.
I was using FF3.
Tried again after restart 11 minutes sharp.

Made a 3000x3000: 6mins 32 seconds

Xeons are optimized for quite different behaviour profiles than the i7. Xeons are made for server situations, and are not specialized for performing a single coordinated task. This relates not only to the CPU (the optimizations there are not so different) but the different Lx caches to which the CPU has access as well as the strategy for reacting to page faults within those caches.

You might well think that that wouldn't make a difference, or not much anyway. Memory is memory. A general rule of thumb, though, is that the ratio of speed difference between Disk Access and RAM access is roughly the same as the ratio between RAM and L2 and the difference between L2 and L1. So multiple orders of magnitude if the work being performed isn't well suited to the caching strategy.

So depending on how the data is localized in RAM, it might be really efficient for the CPU to work on it, while inefficient for another CPU. And if it's not well designed, there can be a huge amount of thrash in server memory no matter which CPU is asked to chew on it.

Just a guess, mind. There might be something else entirely at play.

Gilles you are right that Intel Xeon Processors are optimized and made in a different way for a different market and users, and it would not be good for normal users or for gamers, and they are made specially for servers.

And they perform tasks differently and have different kind and also the inner workings and how is joined all the parts are also better done for a higher perfomance and efficiency.

BUT this is why these Xeon Chips are ALSO used for High End and top Business Workstations where a higher perfomance and speed is needed and more efficiency than the desktop chips can offer, AND you can´t have more than one desktop chip in a computer, BUT you can have 2 or 4 Xeon chips in the same motherboard, or join many special twin motherboards with 2 Xeon to make more complex configurations.

Gilles, I think that what you say is that it depends on the relation, combinationa and communication of hardware, the operative system and how optimized is the software to be able to benefit and get the most from a multi core AND multi chip computer, and is true that this is very important.

COMPARING DIFFERENT COMPUTERS WITH DIFFERENT OS

Quote
GMM For a more precise test you need to perform your benchmark on the same OS.

GMM, yes, you are right that always any kind of speed test or benchmark should be made in the most similar way and be able to compare the computer with the most similar components as this will give much better results to compare more precise and realistically.

I would love to find someone that could have a Windows computer with a dual chip Xeon AND that uses Filter Forge to test it better, BUT regretably I have not found it yet, and I have been searching much through internet, but until now nothing

BUT I still do not understand and can´t explain HOW a computer that is supposed to be a very powerful and fast computer like the MacPro with TWO Chips with 6 cores each (12 cores) at 2,4 GHz each can be SLOWER than just ONE 4 cores chip 3,4 GHz.

I agree with you Gilles D, I will make a research and investigation to find the real truth behind this and HOW works the i7 chips and Xeons and Itaniums, and what is the real difference between them and which one to choose (well surely Itanium not as they cost each one about 3500$ to 4000$ the cheapest one )