Welcome to Fractal Forums

Fractal Software => Mandelbulb 3d => Topic started by: ant123 on February 24, 2011, 01:10:48 PM




Title: A processor benchmark for Mandelbulb 3d
Post by: ant123 on February 24, 2011, 01:10:48 PM
How fast does your computer render this parameter by Theli-at? (render times are displayed at the bottom right corner)
 
  RESULTS:
==================

Celeron M 1.6 ghz
3 minute and 32 seconds

Opteron 154  2.8Ghz (blob)
2 minute and 23 seconds blob, that is slower than a $25 processor from ebay!, but it is a Kool computer.

AMD x2 2.2Ghz (socratease)
2 minute and 10 seconds

1 minute and 49 seconds
Wine on Macbook pro 2.4 C2D (socratease)

Core2Duo 2.0Ghz (ant123)
1 minute and 18.8 seconds

MacBook Pro (2.53 GHz Intel Core i5) under Parallels (KRAFTWERK)
49.2 seconds

Intel Core 2 CPU E8400 @ 3.00 GHz
48 seconds (cytotox)

Intel Core 2 duo @ 3.24 Ghz (E6750)
47 seconds (ant123)

i3 M 350 @ 2.27 GHz (Kameelian)
46.3 seconds

Intel i7-860, Quad Duo Core, 2,8gHz (skyzyk)
38.6 seconds

Phenom 9850 Quad-Core 2.5ghz (Don Whitaker)
38 seconds

Core2Quad 2.4GHz (oppenheimer)
35.8 seconds

Core2Quad Q6600 @2,4GHz (ziegfrid)
34.6 seconds


Core2Quad 2.33GHz (buddhi)
32.3 seconds

AMD Phenom II 965...3.4gh (haltenny)
26.6 seconds

I7 740-QM  @ 1.73GHz (4 Core, 8 Threads, Turboboost: 2.93 Ghz)
25.7 seconds under Linux Wine (ker2x)

i5 750 2.66Ghz (madman)
23.6 seconds

dual quad Xeon 2.0 Ghz, Win-7/64-bit (mrcharles)
22.5 seconds

i7 920 @2.66 Ghz
19.5 seconds (dainbramage)
19.3 seconds (bib)

Phenom X6 3.3ghz (expectation)
18.5 seconds

i7 930 @2.80GHz - Dell XPS (mario837)
17.6 seconds

Intel Xeon X5482 3.2GHz (dual quad cores) 16 GB DDR2 800MHz Vista SP1 64-bit (Nahee_Enterprises)
10.2 seconds

i7 2600k 3.4 GHz [OCd to 4.4 GHz] (4 core/2threads per core), Win7 64-bit (Kameelian)
9.1 seconds

____________________________________
____________________________________
For Mandelbulb3d the computer uses just the CPU and 200-500MB of ram, the gpu is not used, so what matters is the speed of the CPU and the memory, say 1600 memory would be faster than 1333, but you might be able to do larger image renders with more memory.

TEST:
Mandelbulb3Dv16{
M....Uy....u1...S..........2EgavjzHxzqTwzWd7NKzDEEew.GOx.yHJ6BWNLlfaz4ERsIdeBNsD
................................F3j3MMa5nz1........Y./..................y.2.....
.................M.//....6E12...50....EA.....kGlBkqI2ppD/.EvZBcD...E./...w1.BnAH
zEEnAnQD..../2EnAnAnAnAvz0............................................sD...../..
.w1...sDR/dGxbdzTz1fA18BD4gkzeQHfDJaEhyjK90p0REcazH5vZXFiwzrzWhgpmFezzxDWLWLku/.
Uzvzhznth/puzu1BwtvIu8wj......Yx1.....................sD..kz0...................
.............................UJRb4.upRO.sNrd/UkRb4.arRO.cUrd/U9Sb4..............
..........................UX.o2.......0...UD....5/...w2...EG....6/...Mn5...UJI12
XkGf.08cU0AcU0eQ.06Us.Ak.16.2cE.zzzz.MxuvBTuCfuDiUNAT2kYcz1EJzHmWnv4.HEBI.klLS4.
3yhMkskYszfIaQHInDNyzGWfzHx6d2yD2I13.wqN51.e/0biMaOyzcisP8OPAXzD.ezYfTLkRzH.0c..
zzzz.................................2U.8.kzzzD.................................
/6U0.wzzz1...................................QbG6/.cU08.zzFUf74.U08c.szDpZNR..8c
U0EzTFagv0.cU08.QsrRwS4.U08c..bTWmbc..8cU0.lyZLQL/.cU08.Mwbfpe5.U08c.kqTDKLY..8c
U0.kz3fhb0.cU08....cU08.zz/cU08.yz1cU08.xz3cU08..U5QmY1CtM1Oic4Qb/kLWZrLi3KPZB5Q
E....E....E..........6....EGiFLNbJaQU.pPrJaQ.........................c..........
...................3./........zj................................................
................................................................................
.....................2.....3....1....w3Jm3aPnlKMoJ4.............................
.....MU/4MU/4.................zD................................................
................................................................................
................................/....E/...k.....T7pPo34RZ/kHXF5.H/..............
..........U/4M..4MU/4MU/..........................cF.1.......O3E........kz1.....
................................................................................
...........................................2....1....6.....JmZqMj7bP............
.....................................................1........zD................
................................................................................
........................................................}


meanwhile, i found this: http://www.shadoworld.co.uk/UF5/Benchmark/Benchmark%20results.htm


Title: Re: Benchmark your Processor
Post by: Oppenheimer_ on February 24, 2011, 01:27:09 PM
35,8 seconds

Core2Quad Q6600 @2,4GHz, 6gig ram


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: David Makin on February 24, 2011, 03:11:49 PM
Related, 2 benchmarks UPRs for Ultra Fractal with extensive results:

http://www.shadoworld.co.uk/UF5/Benchmark/Benchmark%20results.htm (http://www.shadoworld.co.uk/UF5/Benchmark/Benchmark%20results.htm)

I don't think relative results in Mandelbulb 3D or Mandelbulber will deviate very much from the above, the benchmarks we're actually short of are ones for using GPU rendering - one set for float and hopefully (sometime soon) another set for double :)


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: lenord on February 24, 2011, 04:04:59 PM
Renders it fast enough, what the hell difference does it make. What ever Puter you have is the one you have to use and it's only going to be as fast as it is, this is not a true Benchmark test.


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: ant123 on February 24, 2011, 04:24:02 PM
Len, i took 2 hours and a half rendering a 2mp image last night. I want a faster PC for ASIO/3d/etc. it also gives a comparison across cpu architectures, X4, X6, i7, C2Q, etc.

http://www.shadoworld.co.uk/UF5/Benchmark/Benchmark%20results.htm (http://www.shadoworld.co.uk/UF5/Benchmark/Benchmark%20results.htm)

I don't think relative results in Mandelbulb 3D or Mandelbulber will deviate very much from the above

Thanks, i looked on google for "fractal/sciencemark/pi benchmarks" and i couldnt find a single one! The UF5 comparison is definately the page i was after!



Title: Re: A processor benchmark for Mandelbulb 3d
Post by: blob on February 24, 2011, 05:19:04 PM
2mn 23s on my AMD Opteron 154 (single core 2.8Ghz socket 939 CPU).

 ;D


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: tomot on February 24, 2011, 05:26:06 PM
32.3 sec

quad core cpu q6600 @ 2.4ghz
physX GPU acceleration enabled
4 gb ram



Title: Re: A processor benchmark for Mandelbulb 3d
Post by: Madman on February 24, 2011, 07:29:48 PM
For what it's worth: 23.6 seconds on an i5 750


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: Buddhi on February 24, 2011, 09:30:36 PM
32.8 sec

Intel Core 2 Quad 8200 (2.33GHz)
under Linux - Wine


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: tomot on February 24, 2011, 09:34:55 PM
For what it's worth: 23.6 seconds on an i5 750

thanks for the info. but that's not particularly speedier either.  :o I think I'll wait an 8 core cpu before investing in a new machine.


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: bib on February 24, 2011, 10:44:25 PM
i7 920 @2.66 Ghz => 19.3


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: DarkBeam on February 24, 2011, 10:46:22 PM
A note. Parameters are shared for learning purposes, not for benchmarking. Have respect for the artists use them fairly :)


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: Madman on February 24, 2011, 11:20:12 PM
You're tempting me Darkbeam... I agree on the second part about respect, but in my opinion, there's no rule that says that parameters posted may only be used for a certain purpose. But maybe we should discuss that somewhere in a separate topic...


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: David Makin on February 24, 2011, 11:28:18 PM
A note. Parameters are shared for learning purposes, not for benchmarking. Have respect for the artists use them fairly :)


That doesn't apply to the UF parameters at the link I posted, those were especially designed to cover a large range of variation in actual processing, in particular the second test UPR requires a lot more heavy math in the form of transcendental functions, divides etc.


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: Sockratease on February 25, 2011, 12:17:59 AM
This is an odd file for benchmarking because it doesn't take all that long to render.  When there was a Carrara Benchmark file floating around for a similar unofficial and informal test, it involved many features that were quite time consuming.  The idea was to turn off and on various features in the slowest of the performers, and see what in the program can be blamed.

Anyway - here's what I got  (I turned on Hard Shadows to make it more interesting, but broke down the times accordingly) :

Running under WINE on my MAC  (Macbook Pro - 2.4 GHz Intel Core Duo) :

Main Render; 1 Minute, 14 Seconds
Ambient Shadows; About 3 Seconds
Hard Shadows; 32 Seconds

Total : About 1 Minute 49 Seconds


On my Web Crap Computer  (AMD Athlon 64 Dual Core 4200 @ 2.2 GHz)  (Yes, I have a 64 Bit Processor running a 32 Bit Operating System)  (It was used, and cheap, and it came like that - I plan to upgrade in a decade or 2)  (No graphics card, just the on board chip) :

Main Render; 1 Minute, 31 Seconds
Ambient Shadows; About 3 Seconds
Hard Shadows; 36 Seconds

Total : About 2 Minutes 10 Seconds


On my Offline Computer Computer  (Intel Core 2 Quad Q6600 @2.4GHz, 2.39 GHz) EDIT - Oh yeah, NVidia GeForce 8800 GTS) :

Main Render; 48 Seconds
Ambient Shadows; About 2 Seconds
Hard Shadows; 13 Seconds

Total : About 1 Minute 3 Seconds


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: dainbramage on February 25, 2011, 01:01:11 AM
My system is the same or similar to bib's, took 19.5 seconds.


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: mrcharles on February 25, 2011, 05:34:11 AM
22.5 sec. for total render (incl. shadows)

dell 690 workstation dual quad xeon 2.0 Ghz, win7/64


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: ker2x on February 25, 2011, 09:48:46 AM
how to load the benchmark please ?  :'(

Edit : Oops ! It's not mandelbulber :)


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: KRAFTWERK on February 25, 2011, 12:31:17 PM
49.2 seconds on a MacBook Pro (2.53 GHz Intel Core i5) under Parallels running XP... :P

I have to get a new puter!  O0


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: ker2x on February 25, 2011, 01:37:16 PM
25.7s on my laptop with wine on linux (ubuntu 64bits)

CPU : I7 740-QM  @ 1.73GHz  http://ark.intel.com/Product.aspx?id=49024
(4 Core, 8 Threads, Max Turbo Frequency : 2.93 Ghz)


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: haltenny on February 25, 2011, 04:12:20 PM
26.6 seconds on AMD Phenom II 965... (3.4gh)


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: cytotox on February 25, 2011, 06:05:26 PM
48 sec

Intel Core 2 CPU E8400 @ 3.00 GHz, 8 Gb RAM
Windows Vista 64 Bit, SP2


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: ant123 on February 26, 2011, 04:17:19 PM
Thanks for all the replies! i have written them in the results at the top! Moor's law says that the speed should be 10 times faster every 60 months / 5-6 years!

On my Offline Computer Computer  (Intel Core 2 Quad Q6600 @2.4GHz, 2.39 GHz) EDIT - Oh yeah, NVidia GeForce 8800 GTS) :

Main Render; 48 Seconds
Ambient Shadows; About 2 Seconds
Hard Shadows; 13 Seconds

Total : About 1 Minute 3 Seconds



that is abotu as fast as me on my c2 duo 2.0 ghz, either there was another process running or it was only on 2 cores.


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: ant123 on March 12, 2011, 07:27:59 PM
I don't think relative results in Mandelbulb 3D or Mandelbulber will deviate very much from the above,

the phenom 3.4/i7 920 gets 26/19 on this parameter and 19/13 on the other test, i dont know if it does the same on all parameters, else perhaps the phenom ii has extra cash and sse instructions compared to the athlon ii to make it 20+ % better at these maths.


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: skyzyk on March 13, 2011, 03:18:11 AM
38.6 Seconds

Intel i7-860, Quad Duo Core, 2,8MHz


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: Don Whitaker on March 13, 2011, 04:10:50 AM
about 38 seconds

Processor: AMD Phenom  9850 Quad-Core , 2500 Mhz
Windows 7 32 bit, 3 GB RAM

I've been scheming ways to afford a system just for rendering  :pray2:  It's nice to have some rough estimates of the boost I'd get from upgrading.


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: mario837 on March 15, 2011, 02:46:55 PM
Dell studio XPS 9100 with i7 930 @2.80GHz and 4.0Gb RAM = 17.6 seconds


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: ant123 on March 17, 2011, 02:53:37 AM
thanks for those, 2.8 ghz is standard for i7 930  i think...

if so, a stock i7 2600k would do around 15.5 seconds and the same OC'd to 4.8 could be around 10s.

http://www.bjorn3d.com/Material/revimages/cpu/Core_i7_2600K/Sandra_Processor_Multimedia1.jpg


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: Xenodimensional on April 11, 2011, 02:49:17 PM
6.5 seconds  :o

dual xeon x5650
24GB ram


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: heavenriver on April 11, 2011, 04:57:04 PM
Don't know if you still need benchmarks, but here's mine!

Processor: Intel Core i5 750 @ 2.67 Ghz
4GB DDR3 RAM

Total rendering time: 24.3 seconds


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: ziegfrid on April 19, 2011, 04:02:33 PM
34.6 seconds

Core2Quad Q6600 @2,4GHz, 4GB ram


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: kameelian on August 02, 2011, 06:20:32 PM
for what it's worth, this late on

i3 M 350 @ 2.27 GHz, (twin core/2threads per core), 3GB RAM, Win7 Home Pre, 32bit. general laptop. = 46.3secs plus 1.2 for shadow.
i7 2600k 3.4 GHz [OCd to 4.4 GHz] (4 core/2threads per core), 8GB RAM, Win7 Home Pre, 64bit. Renderer Only. = 9secs plus .1 for shadow.
Kam


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: ant123 on September 04, 2011, 11:23:27 PM
thanks Kameelian, it's logged.

that is realy fast.


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: kameelian on September 05, 2011, 10:11:04 AM
Hi,

Hah, yeah, I suppose it is quite fast (compared to some).

But it is still really slow - waiting, waiting.

cheers
Kam


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: cKleinhuis on October 09, 2013, 08:44:18 AM
i7 4770k 3,5 ghz - ~11 Seconds :D
(damn, i love that machine!)


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: Stonevoice on October 15, 2013, 07:23:43 PM
It took me 3:57.
My computer is nine years old, with an AMD Athlon 3000+ proc with 1.43 GB Ram.
You guys are lucky!

My computer also won't let me run any formulas with dIFS in them, because of the graphics card.


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: cKleinhuis on October 15, 2013, 07:38:44 PM
that is an really old computer :(
but for your difs problem i can help you :)
you cannot combine difs with "normal" functions, you find compiler messages in the main window below the tab area,
you can just combine difs formulas with other difs functions ;) mb3d is in no way using a gpu!


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: Stonevoice on October 15, 2013, 07:54:03 PM
Thanks, cKleinhuas, but I cannot run dIFS  the 'right' way either -in other words, just dIFS combined with dIFS. The formula box does not even show the 'dIFS' headings at among the formula suggestions. The list ends with the 'Ads' column.
If I try to run someone esles params that use dIFS, I get an error message saying that the processor does not have the right instruction set to run those formulas.
I also have a laptop that is about as old, and it CAN run dIFS - I guess it has different CPU or GPU in it.
oh well, getting a new computer in a week with i7 3.5 mHz with 8 GB RAM, so things will change. ^-^

I hope I can run MB3D on 64 bit system...


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: cKleinhuis on October 15, 2013, 08:16:57 PM
it will work perfectly on that ;)
the cpu might be really missing some features that darkbeam hacked into those formulas @darkbeam, no offence you are doing a fantastic job!


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: DarkBeam on October 18, 2013, 04:54:55 PM
 :D Not really - I just use a few of the oldest CPU features. Just floating point routines... And some xor/and/jg that were also in z80 machines I think ;D


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: fractalgee on July 23, 2015, 04:48:40 PM
Dual quad core Xeon Mac Pro @2.33GHz

14.5 seconds including ambient shadows calc


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: xahhax on January 23, 2016, 01:19:28 PM
2015 iMac 5K Skylake: 0:09.1


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: AtomicNixon on November 28, 2017, 04:03:28 AM
Threadripper 1950x at 4.1 Ghz.  Two point eight seconds.  :D  I win! 


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: hobold on November 28, 2017, 01:31:06 PM
Threadripper 1950x at 4.1 Ghz.  Two point eight seconds.  :D  I win! 
The name "AMD Ryzen Threadripper" is pretty much the only downside of that monster :). Even the price looks good if you compare to Intel - especially if you want to use ECC RAM to help with system stability for those week long renders (support for ECC RAM requires even more expensive Xeons on the Intel side of the fence). Performance is very competitive for anything but AVX code, and even there AMD ain't bad.

(Disclaimer: I am now officially an AMD fanboy, since I built a personal supercomputer based on threadripper. Not overclocking, though; I went for reliability and low noise on air cooling.)

Competition is a good thing. Nice to have it back.


Title: Re: A processor benchmark for Mandelbulb 3d
Post by: AtomicNixon on December 10, 2017, 09:44:52 PM
The name "AMD Ryzen Threadripper" is pretty much the only downside of that monster :). Even the price looks good if you compare to Intel - especially if you want to use ECC RAM to help with system stability for those week long renders (support for ECC RAM requires even more expensive Xeons on the Intel side of the fence). Performance is very competitive for anything but AVX code, and even there AMD ain't bad.

(Disclaimer: I am now officially an AMD fanboy, since I built a personal supercomputer based on threadripper. Not overclocking, though; I went for reliability and low noise on air cooling.)

Competition is a good thing. Nice to have it back.

Thanks!  And congrats on the new machine!  If you want to use all of it, go for the Enermax full block cover cooler.  I'm using the 360mm model and can add my voice to others in testifying that it's the single best cooler on the market by a wide margin.  Full coverage on the package counts for about 10 degrees C which makes sense given that the dies are situated in the corners.  Most coolers are still trying to cover a square with a circle, not good.