Benchmarking Jitter's CPU/GPU performance on your computer
I am starting a new thread with Pedro Santos's nicely crafted Jitter benchmark patch.
It is taken from my post titled "best computer (graphic card etc..) for Jitter".
This more important part was buried to deep in the thread to be missed out on.
Here is his reply with the patch "jitter_benchmark-v1.01.maxpat" :
(this patch version has "cache_mode immediate" to the jit.gl.gridshape object present in the GPU Geometry 1)
Hello, again! Here’s a better Jitter benchmark utility I just built. It could be a start for a more definitive tool with the help of the community.
It’s also theoretically expandable to include other scenes…
For your reference here’s the data for my custom-built desktop:
CPU: 97.8
GPU Geometry 1: 110.8
GPU Geometry 2: 98.3
GPU Pixel Shaders: 96.5
Windows 7 (64 bit), Max 6.1.9 (32 bit)
Intel Core 2 Quad Q9400 @ 3.2 GHz, 6GB RAM
AMD HD 4870 1GB RAM
I didn't see this thread earlier
Lenovo w520
2.2ghz cpu
16gb ram
Nvidia quadro 2000m
max 6
CPU: 117.7
GPU Geometry 1: 55.1
GPU Geometry 2: 363.4
GPU Pixel Shaders: 28.4
max 7
CPU: 150.9
GPU Geometry 1: 37.1
GPU Geometry 2: 359.7
GPU Pixel Shaders: 44.2
MAX 7 test patch version 1.012
CPU: 153.8
GPU Geometry 1: 36.3
GPU Geometry 2: 200.8 I didn't see the background during the test.
GPU Pixel Shaders: 39.4 This test was just a bunch of frozen noise
Geometry test 1 uses jit.gl.multiple.
Geometry test 2 uses 3 high poly-count objects.
Professional graphics card drivers are usually optimized for these kinds of operation (modelling with many polygons).
On the other hand, the shader performance is low.
It's really useful to see such differences between cards...
macbook pro (from 4 5 years back)
OSX 10.9.5
2.2 ghz intel icore i7
8 GB 1333 Mhz DDR3
AMD Radeon HD 6750M 1024 MB
CPU: 96.2
GPU Geometry 1: 20.4
GPU Geometry 2: 96.8
GPU Pixel Shaders: 47.2
Results from three MacBook Pros of various ages. Disappointing numbers from the brand new lappy...
MacBook Pro 15" Early 2008
OSX 10.6.8
Max 6.0.7
2.6GHz Intel Core Duo
2GB 667MHz DDR2 SDRAM
NVIDIA GeForce 8600M GT
CPU: 59.4
GPU Geometry 1: 5.8
GPU Geometry 2: 7.8
GPU Pixel Shaders: 19.4
—————
MacBook Pro 15" Late 2011
OSX 10.7.5
Max 6.1.8
2.5GHz Intel Core i7
4GB 1333Mhz DDR3
AMD Radeon HD 6770M 1024MB
CPU: 93.1
GPU Geometry 1: 19.8
GPU Geometry 2: 95.4
GPU Pixel Shaders: 39.3
—————
MacBook Pro 15" Retina Mid 2014
OSX 10.9.5
Max 6.1.8
2.5GHz Intel Core i7
16GB 1600MHz DDR3
NVIDIA GeForce GT 750M 2048MB
CPU: 109.9
GPU Geometry 1: 6.4
GPU Geometry 2: 27.0
GPU Pixel Shaders: 84.1
MacBook Pro 15" non Retina Mid 2012
OSX 10.9.4
Max 6.1.7
2.6GHz Intel Core i7
8GB 1600MHz DDR3
NVIDIA GeForce GT 650M 1024MB
CPU: 108.6
GPU Geometry 1: 5.2
GPU Geometry 2: 22.6
GPU Pixel Shaders: 77.4
MacBook Pro
Retina, 15-inch, Late 2013
2,6 GHz Intel Core i7
16 GB 1600 MHz DDR3
NVIDIA GeForce GT 750M 2048 MB
CPU: 126.2
GPU Geometry 1: 6.4
GPU Geometry 2: 26.8
GPU Pixel Shaders: 82.5
what a disappointment...
Macbook pro 5.3
15" mi-2009
2.8Ghz Core 2 Duo
8GB RAM
MAX 6
NVIDIA GeForce 9400M 256 MB
CPU: 36.7
GPU Geometry 1: 2.7
GPU Geometry 2: 16.9
GPU Pixel Shaders: 7.7
NVIDIA GeForce 9600M GT 512 MB
CPU: 49.1
GPU Geometry 1: 4.6
GPU Geometry 2: 50.3
GPU Pixel Shaders: 21.9
MAX 7
NVIDIA GeForce 9600M GT 512 MB
CPU: 61.7
GPU Geometry 1: 6.6
GPU Geometry 2: 53.4
GPU Pixel Shaders: 25.7
mbp 15" (4 days old)
OSX 10.9.4
Processor 2.8 GHz Intel Core i7
Memory 16 GB 1600 MHz DDR3
Graphics NVIDIA GeForce GT 750M 2048 MB
1st version
CPU: 109.8
GPU Geometry 1: 6.3
GPU Geometry 2: 26.5
GPU Pixel Shaders: 82.9
2nd version
CPU: 109.6
GPU Geometry 1: 6.4
GPU Geometry 2: 27.2
GPU Pixel Shaders: 84.9
not much difference
Hackintosh OSX.9.5 I7-2700K @ 4.5 GHz 16 GB 1600 MHz RAM AMD6870
CPU: 110.4
GPU Geometry 1: 31.0
GPU Geometry 2: 112.7
GPU Pixel Shaders: 168.8
iMac 27" (late 2012),
OS X 10.8.5
3.4 GHz Intel Core i7
8GB RAM
NVIDIA GeForce GTX 675MX (1024 MB):
Max 6.1.5
CPU: 107.7
GPU Geometry 1: 6.8
GPU Geometry 2: 32.1
GPU Pixel Shaders: 207.9
Macbook Pro (late 2013)
OS X 10.9.5,
2.6 GHz Intel Core i7
16GB RAM
NVIDIA GeForce GT 750M (2048 MB)
Max 6.1.5
CPU: 110.3
GPU Geometry 1: 4.9
GPU Geometry 2: 27.2
GPU Pixel Shaders: 85.6
Boy looking at some of your guys results , I can’t help but have the feeling that I washed 3334$ canadian dollars down the drain with this week old 15" mbp.
I've built this crazy jitter/physic/kinect patch for a show i'm doing next friday and it stutters like crazy.
I guess I'll have to face the fact that i'll have to go windows pc for jitter from now on.
thanks
Hello. ZIP BOTERBLOEM, since you've got a Hackintosh, could you run the test on a Windows 7 or 8 partition (if you have one..)?
That would be some useful information for comparison purposes.
And we don't have many AMD results in Windows.
It is very strange that my geometry results using an "old" AMD HD4870 1GB in Windows are better than almost anyone with better and newer cards. Coincidentally, under Windows, even a relatively new NVIDIA card (example: GTX670 2GB) got much worse results. Compare:
Pedro Santos's Desktop System:
Windows 7 (64 bit), Max 6.1.9 (32 bit)
Intel Core 2 Quad Q9400 @ 3.2 GHz, 6GB RAM
AMD HD 4870 1GB RAM
CPU: 97.8
GPU Geometry 1: 110.8
GPU Geometry 2: 98.3
GPU Pixel Shaders: 96.5
DTR's Desktop system:
i5 4690k 3.5GHz
GA-Z87X-OC mobo
EVGA (Nvidia) GTX670 2GB FTW gfx card
8GB RAM 1600Mhz
Win 7 pro 64 bit
Max 6.1.9 32bit
CPU: 157.9
GPU Geometry 1: 10.6
GPU Geometry 2: 54.3
GPU Pixel Shaders: 245.8
DTR's computer is theoretically better than mine in every aspect. Both are running the same OS. The difference is he has better and newer graphics card but from another vendor (NVIDIA). Could it be that AMD's Windows drivers for OpenGL are more optimized than NVIDIA's for geometry drawing?
So, I think we would benefit from trying to understand these performance problems. Possibilities for this:
- AMD vs NVIDIA drivers
- Windows vs MAC OS X driver optimizations
- a combination of the previous two.
- Older vs Newer cards' Jitter optimization (or lack of)? Sometimes, modern graphics cards have different ways of doing things. Since Jitter's OpenGl implementation is not very modern (I guess it might be OpenGL 2.. I'm certain it isn't OpenGL 4...), these performance boosts might not be exposed to Jitter. This is pure speculation on my part...
By the way, don't be sad because of the very low geometry numbers! ;-)
They might mean very little in a practical context because of the very high number of polygons being drawn.
If you open the scenes' sub-patches and change them, the performance will certainly quickly ramp up.
In geometry1: change the dim of the jit.gl.gridshape to something like 24 24; also change the number of instances of jit.gl.multiple being drawn by changing the dim of jit.noise.
In geometry2: change the dim of the jit.gl.gridshapes from 320 320 to something less extreme...
Since I did the benchmark in my machine, the resolution and processing involved in each scene was done with the goal to achieve 100 fps, so it's all relative to that reference... anyway i think it would be useful to understand these very pronounced differences performance-wise...
One last thing: don't forget that Jitter is not multi-threaded... so for this purpose your shiny eight-core won't be so much superior than a dual-core right now... I'm crossing my fingers to see that changed in Max 7... by the way, how cool would it be if someone from Cycling just to mess with us run this benchmark and presented their results with huge gains when compared with Max 6? ;-)
Pedro
P.S: I'm starting to get worried about Sam. Has anyone seen him lately? The last time we heard from him, he was going out with Scarlett Johansson... it's been what... two weeks?
Hola todos,
even I don't do a lot of work in Jitter I think it is a great thread with all the informations. I thought maybe all the benchmarking results could be collected in an overview table as a wiki page for easy comparison. Would start doing myself but very litle time these days...
What do you think?
Cheers Jan.
@Pedro: Give him some private time with Scarlett. The first weeks with a new love are precious!! He'll be back I am sure :)
Yes I do hope things change in max 7 for a multi-threaded Jitter. Take advantage of the machine we have :-)
Thanks again Pedro
phiol
Hi Jan.
I was thinking of something even more automated, but never tried it myself: upload the results via php queries to a MySQL database on the web.
I've some minimal experience with it, but never tried to do it with Max however. If someone has some experience with this, it would be great.
But your solution is a great alternative!
if there is the backend you can simply do it with maxurl and a post request. but there should be a login or some other meassures to prevent spam. further more there is work to do to collect the hardware data ..
for simplicity i'd go for the wiki. also the results will stay on the c74 site where new people will most likely search first.
Hello again, guys.
Good points, Jan. I agree with your that the results should be located in Cycling'74 website, so I guess the Wiki should be the best solution.
But before someone starts gathering info in the Wiki, I think it would be useful to improve the benchmark to a more definitive status.
For example, given the discrepancies in the results, I think it would be useful to have a baseline value based on a more conventional machine, that better represents the majority of Max/Jitter users. From the results up until now, it seems to be a MacBook Pro with a discrete graphics card (Nvidia or AMD). Any thoughts?
And someone should also build a nice physics test!
Any other tests suggestions? Jitter with audio processing on?
MacBook Pro
Retina, 15-inch, Late 2013
2,3 GHz Intel Core i7
16 GB 1600 MHz DDR3
NVIDIA GeForce GT 750M 2048 MB
CPU: 104.9
GPU Geometry 1: 6.4
GPU Geometry 2: 26.9
GPU Pixel Shaders: 99.3
... and don't forget to mention your computer specs!"
First: Great idea and thanks for the test patch!
Self build PC:
i5 3570k itx board, no Overclocking
8 GByte RAM
FirePro W600 1x 1280x1024 + 4x 1920x1080 simultaneously connected
WIN7 Pro 64bit , Max 6.1.8 32bit
Slightly different results depending on which screen the test is running
CPU: 143.6
GPU Geometry 1: 77.7
GPU Geometry 2: 238.1
GPU Pixel Shaders: 77.9
I was working for years on Macs but I had to switch to PC ... (see "trash can" values below)
There is currently only one big fault I have with crashes on quit with WIN7 standalones I already posted in this thread:
It would be nice if one of you guys could verify that too (or better, you have a solution for that, C74 has no idea).
Thanks and sorry if that is too cross posted, but nobody has an idea about that since months...
Hi guys ('n girls?)
I'm back from touring. Will do the test on my hackintosh that runs both Win7 and OSX10.8 as well, for OS comparison. Interestingly on that computer the 2nd part of the benchmark hangs OSX entirely and crashes Max in Win7. I need to look into this before reporting. The computer has no network/internet at the moment and I have 2 more shows this week. Will do ASAHP (= as soon as humanly possible).
I also have an old AMD HD6870 gfx card lying around somewhere so I'll swap out my Nvidia GTX670 for direct AMD-Nvidia comparison. Not this week though.
D.
Sorry, don't have a copy of Windows running on my Hack.
From the hack world I know that AMD and NVIDIA both have their specific strengths. AMD is quite good at OpenGL.
I expect a modern card like a AMD R9 280X to be three times as fast as my 6870 for OpenGL.
The old axe....
MacBook Pro (17-inch, Mid 2010)
2.66 GHz Intel Core i7
8 GB 1067 MHz DDR3
NVIDIA GeForce GT 330M 512 MB
OS X 10.10
Max 6.1.9 64bit
CPU: 59.9
GPU Geometry 1: 6.4
GPU Geometry 2: 29.2
GPU Pixel Shaders: 28.4
well cheer up
your old axe performs farely equal to my 2 week old newborn mbp
:-)
phiol
And the trash can...
Modellname: Mac Pro
Modell-Identifizierung: MacPro6,1
Prozessortyp: Quad-Core Intel Xeon E5
Prozessorgeschwindigkeit: 3,7 GHz
Anzahl der Prozessoren: 1
Gesamtanzahl der Kerne: 4
L2-Cache (pro Kern): 256 KB
L3-Cache: 10 MB
Speicher: 12 GB
OSX: 10.10
CPU: 116.0
GPU Geometry 1: 25.7
GPU Geometry 2: 163.6
GPU Pixel Shaders: 192.7
CPU: 59.2
GPU Geometry 1: 10.6
GPU Geometry 2: 52.7
GPU Pixel Shaders: 15.8
window as it comes.
macbook air Mid 2011 / 4GB memory
Howdie, I 'm doing some more testing. My i5 GTX670 system results were already posted:
i5 4690k 3.5GHz
GA-Z87X-OC mobo
EVGA (Nvidia) GTX670 2GB FTW gfx card
8GB RAM 1600Mhz
Win 7 pro 64 bit
Max 6.1.9 32bit
CPU: 157.9
GPU Geometry 1: 10.6
GPU Geometry 2: 54.3
GPU Pixel Shaders: 245.8
I now changed the geometry dim's in GPU geometry 1 and 2 to get close to Pedro's 100 score baseline.
GPU Geometry 1: 99.9 > dim 11 11
GPU Geometry 2: 105.6 > dim 220 220
That's a whole lot lower than the original settings.
Now if I can manage to find that old AMD HD6870 card I'll swap it out and do a direct comparison...
Found it... The good ol' HD6870 destroys the shiny new GTX670 (ok, not so new, but still...).
i5 4690k 3.5GHz
GA-Z87X-OC mobo
XFX (AMD) RADEON HD6870 2GB gfx card
8GB RAM 1600Mhz
Win 7 pro 64 bit
Max 6.1.9 32bit
CPU: 124.2 > this one's funny as we 'd expect the CPU test not to be much different with another GPU, still there's a 35fps drop
GPU Geometry 1: 185.5
GPU Geometry 2: 193.9
GPU Pixel Shaders: 201.5
(tested at original dim settings)
Old AMD cards beating newer Nvidia ones by so much shouldn't be the case. In 3d engine benchmarks like Heaven (in openGL mode) the GTX670 scores multiples of what the HD6870 does. Anyone at C74 willing to share some insights?
intel core i5-4590 3.30ghz
GeForce GTX 750 ti
16GB RAM
win 8.1
Max 6.1.9
CPU: 158.1
GPU Geometry 1: 8.5
GPU Geometry 2: 34.4
GPU Pixel Shaders: 172.6
This computer is only a month or two old and cost a bit - is it just me, or does it seem to be producing some pretty crap figures here? if so, any ideas? (im fairly new to this, and i'd bought this computer exclusively to use Jitter)
@daddymax: If you care to read the thread above you should get some answers. Apart from that, keep in mind that this is a synthetic test with unusual parameters on the geometries. In your real life applications you might get fine results. I get equally bad GPU results in this test while in my actual projects I run at over 100fps steady.
These results are from standalones on a Lenovo ideapad Flex 14 Hackintosh, running OSX 10.9/Max 6.18(32bit) and WIN 8.1(64bit)/Max 6.19(32bit).
It uses the built in HD4400 GPU.
i5 4200U, 4 GByte , SSD(OSX), SSHD(WIN)
Screen is 1366x768, I assume this lifts the values somehow...
Unfortunately I had not the same Max versions.
OSX Hackintosh :
CPU: 60.9
GPU Geometry 1: 13.6
GPU Geometry 2: 72.2
GPU Pixel Shaders: 24.0
I got much better results when hiding the dock (had no secondary monitor at place to check on it)
WIN 8.1:
CPU: 81.6
GPU Geometry 1: 4.2
GPU Geometry 2: 84.0
GPU Pixel Shaders: 27.8
Hello.
Welcome back, DTR! Your AMD HD6870 results confirmed what I was suspecting: AMD's OpenGL drivers, at least on Windows, seem to be more optimised than NVidia's. I suspect that NVidia's OpenGL drivers might be intentionally limited at least on high polygon geometry drawing, typically used on professional CAD markets, in order to sell professional Quadro graphics cards instead of consumer/gaming Geforce cards...
We have some kind of proof on this with JNINEK's results:
Lenovo w520
2.2ghz cpu
16gb ram
Nvidia quadro 2000m
CPU: 117.7
GPU Geometry 1: 55.1
GPU Geometry 2: 363.4
GPU Pixel Shaders: 28.4
BENE's computer results, using a professional AMD graphics card, also shows that the drivers are optimized for high poly drawing (geometry test 2). The difference here is that the regular AMD gaming card's drivers are not so limited in this regard.
Self build PC:
i5 3570k itx board, no Overclocking
8 GByte RAM
FirePro W600 1x 1280×1024 + 4x 1920×1080 simultaneously connected
WIN7 Pro 64bit , Max 6.1.8 32bit
Slightly different results depending on which screen the test is running
CPU: 143.6
GPU Geometry 1: 77.7
GPU Geometry 2: 238.1
GPU Pixel Shaders: 77.9
I would love to see the results of a modern AMD graphics card (R9 280 or R9 290, or similar).
@Pedro: Interesting theory. I'll see if I can do some more tests in the coming days to check if there is a clear poly count threshold where the Nvidia's drop off. If this is really the case it might be useful to add one or more tests to the benchmark that have a poly count closer to what one would encounter in practical projects.
Interesting results...
2013 Macbook Air 11" i5 1.3ghz
CPU: 69.4
GPU Geometry 1: 23.2
GPU Geometry 2: 106.6
GPU Pixel Shaders: 42.7
Intel HD5000
2013 Toshiba Qosmio i7 2.4ghz
CPU 139.7
GPU Geometry 1: 4.1
GPU Geometry 2: 18.6
GPU Pixel Shaders 122.3
NVIDIA Geforce GTX670M
SECRETKILLEROFNAMES, very interesting indeed! (in a strange kind of way ;-)
To me, it really seems like a driver implementation issue.
Intel HD5000 around 5 times better than a GTX670M? That's crazy.
This thread's subject is starting to become more relevant, given these odd results...
Cycling, Rob Ramirez, any possible explanations for these results?
Hardware architecture differences? Driver implementation decisions?
Jitter design decisions favoring a certain kind of OpenGL method / implementation?
I've said before that these geometry scenes have a very high poly-count, so they might not be your typical everyday scenario.
In a couple of days, I could post an update on the benchmark patch that tests each scene while gradually increasing the geometry and pixel's dim values, in order to obtain a more complete opinion of the tested hardware.
Anyway, given the radical differences seen, perhaps Cycling'74 should have some sort of recommendation on the subject? Do you guys ever noticed such performance differences on different brands (AMD, NVIDIA, Intel)?
MacBook Pro (Retina, 15-inch, Late 2013)
2 GHz Intel Core i7 - 16 GB 1600 MHz DDR3
Intel Iris Pro 1536 MB
OSX 10.10
Max 6.1.9 32bit
CPU: 95.2
GPU Geometry 1: 31.7
GPU Geometry 2: 125.9
GPU Pixel Shaders: 65.2
Yep, those integrated graphics laptops outdoing GTX cards nearly as big as those laptops is beyond ridiculous :D
I filed a support request on the subject to C74 yesterday. We should hear about it soon.
Just had a thought: anyone capable of setting up a quick similar high poly scene in Processing or other openGL environment to compare? Not a full benchmark, just one scene. Curious whether we 'd see the same slow down on Nvidia cards.
On a somewhat related note, here's an interesting article from Tom's Hardware comparing professional cards with gaming cards in a variety of benchmarks:
http://www.tomshardware.co.uk/best-workstation-graphics-card,review-32728.html
Hi and thank you for the test.
Macbook Pro mld-2009 15"
2.8 Ghz Intel Core 2 Duo
4 Go 1067 Mhz DDR3
Max 6.1.8
10.9.5
NVIDIA GeForce 9600M GT 512 Mo
CPU: 53.9
GPU Geometry 1: 6.3
GPU Geometry 2: 51.3
GPU Pixel Shaders: 25.6
NVIDIA GeForce 9400M 256 Mo
CPU: 51.0
GPU Geometry 1: 2.7
GPU Geometry 2: 17.5
GPU Pixel Shaders: 8.0
For fun I ran it on my 7 years old macbookpro 2,2:
Core2duo 2.33GHZ
2GB RAM
ATI Radeon X1600 256MB
OSX 10.7.5
Max 6.1.2 (oops seems I need to update)
CPU: 46.7
GPU Geometry 1: 3.7
GPU Geometry 2: 25.5
GPU Pixel Shaders: 6.3
Just to add to the database:
Model Name: MacBook Pro
Model Identifier: MacBookPro8,2
Processor Name: Intel Core i7
Processor Speed: 2.2 GHz
Number of Processors: 1
Total Number of Cores: 4
L2 Cache (per Core): 256 KB
L3 Cache: 6 MB
Memory: 8 GB
PCI: AMD Radeon HD 6750M
built in: Intel HD Graphics 3000
Max 6.1.7
OSX 10.6.8
Note variation in results w 2 consecutive tests:
1st time:
CPU: 99.6
GPU Geometry 1: 19.6
GPU Geometry 2: 74.7
GPU Pixel Shaders: 39.0
again:
CPU: 118.4
GPU Geometry 1: 19.5
GPU Geometry 2: 79.7
GPU Pixel Shaders: 44.9
Model Name: Mac Pro
Model Identifier: MacPro6,1
Processor Name: 6-Core Intel Xeon E5
Processor Speed: 3,5 GHz
Number of Processors: 1
Total Number of Cores: 6
L2 Cache (per Core): 256 KB
L3 Cache: 12 MB
Memory: 16 GB
Graphics Card: 2x AMD FirePro D500
Bus: PCIe
VRAM: 3072 MB each
Max: 6.1.7
OSX: 10.9.5
CPU: 109.3
GPU Geometry 1: 21.4
GPU Geometry 2: 108.7
GPU Pixel Shaders: 169.2
Just tried the test with M7. Results are the same as for M6 on my Hack.
So is the verdict in general that Jitter in Max7 performs the same (multithread support or not)?
General verdict based on 1 Max7 test...? I'll try 'n benchmark my system tomorrow.
I haven't been keeping up with the other threads, so i'm wondering if just anecdotally if people are noticing for example : more stable video playback in jitter, but my apologies if this thread isn't the appropriate place to gauge that
Mac osx 10.6.8 ; max 6.1.9
Identifiant du modèle : MacBookPro6,2
Nom du processeur : Intel Core i7
Vitesse du processeur : 2,66 GHz
Nombre de processeurs : 1
Nombre total de cœurs : 2
Cache de niveau 2 (par cœur) : 256 Ko
Cache de niveau 3 : 4 Mo
Mémoire : 4 Go
(sry, too lazy to translate, should not be too difficult to understand surely ?...)
CPU: 75.7
GPU Geometry 1: 7.5
GPU Geometry 2: 55.7
GPU Pixel Shaders: 29.1
...soo, those numbers are FPS right ? so, the higher the better right ? (sorry to sound so stupid.)
@DTR: In my experience, Max 7 is performing exactly the same as Max 6. I guess the substantial performance improvements in Jitter are solely related to video performance (better use of the GPU in the video decoding department).
But I would like to know if the "Jitter world" (not to be confused with jit.world!) is still being processed in the main (low-priority) thread, sharing resources with, for instance UI drawing and other stuff, and if it's possible to make use of multiple processors with Jitter (multi-threading). But perhaps this is a question for another post, I'll open a new topic for this.
@AWEPITTANCE: The video playback improvements, for now, are Mac OS X only. A solution for Windows is being developed and will be introduced in a later update.
@VICHUG: Yes, those number are frames per second (total number of frames drawn dividing by the number of seconds the test runs).
@Pedro: I'm getting large improvements on the CPU and shader test. GPU geometry is still lousy though.
Max 7.0.0 32bit >
CPU: 199.7
GPU Geometry 1: 6.9
GPU Geometry 2: 54.6
GPU Pixel Shaders: 329.5
Max 6.1.9 32bit >
CPU: 165.4 (edited: i had a strangely low score here the first time round, running it again restored it)
GPU Geometry 1: 10.7
GPU Geometry 2: 54.9
GPU Pixel Shaders: 245.8
i5 4690k 3.5GHz (slight OC)
GA-Z87X-OC mobo
EVGA (Nvidia) GTX670 2GB FTW gfx card
8GB RAM 1600Mhz
Win 7 pro 64 bit
And here's something else funny. The benchmark crashes my other system when it starts GPU geometry 1. It's a hackintosh with OSX and Win7. On OSX is freezes the screen except for the mouse. Needs a hard reset. On Win7 Max crashes and the attached error appears. This is the link in the error message: http://nvidia.custhelp.com/app/answers/detail/a_id/3007
It happens with both Max 6 and 7. Seems like the graphics card can't handle the amount of poly's thrown at it.
i7 2600k
Nvidia GTX 660Ti (MSI brand)
GA-Z68MA-D2H-B3 mobo
8GB 1600MHz RAM
Win7 pro 64bit & OSX 10.8.5
It might be something specific to this build. I stopped using it for performances because when under medium load it creates lots of audio drop outs, although the system specs should be up to the task. Perhaps crappy motherboard or power supply at fault?
intel core i5-4590 3.30ghz
GeForce GTX 750 ti
16GB RAM
win 8.1
Max 7
CPU: 197.3
GPU Geometry 1: 7.7
GPU Geometry 2: 34.4
GPU Pixel Shaders: 173.4
look at that - i posted earlier stats using 32 bit max 6 - using 64 bit max 7 all figures are similar, but CPU has gone from 158.1 to 197.3 - not too shabby.
I have a Hackintosh with OS X 10.10 Yosemite installed. No Windows partition for comparison (sorry guys). Right now I'm looking to buy a new graphics card (R9 290x sounds great or Nvidia GTX 970) but this thread is making me want to wait until we get more information.
i7 4770k
16GB RAM
Intel HD 4600
Samsung Evo 840 SSD
Max 7 64bit
CPU: 221.0
GPU Geometry 1: 20.0
GPU Geometry 2: 122.3
GPU Pixel Shaders: 35.1
Max7 32bit
CPU: 176.1
GPU Geometry 1: 19.8
GPU Geometry 2: 120.5
GPU Pixel Shaders: 35.0
Max 6.1.9 64bit
CPU: 189.2
GPU Geometry 1: 19.9
GPU Geometry 2: 110.6
GPU Pixel Shaders: 34.6
Max 6.1.9 32bit
CPU: 117.5
GPU Geometry 1: 19.6
GPU Geometry 2: 110.4
GPU Pixel Shaders: 34.4
15-inch, Early 2011 Macbook Pro
2,2 GHz Intel Core i7 (quadcore)
16 GB 1333 MHz DDR3
AMD Radeon HD 6750M 1024 MB
256 GB SSD
OS X 10.8.5
Max 7
CPU: 121.9
GPU Geometry 1: 19.7
GPU Geometry 2: 115.1
GPU Pixel Shaders: 47.4
@Pedro: ...I would love to see the results of a modern AMD graphics card (R9 280 or R9 290, or similar).
Here it is, it also shows a big CPU/Geometry1 difference between Max6 and 7 (but wow, this combination rocks):
Self build PC
XEON E3 1241v3 (3.5/3.9GHz Quadcore)
AMD R9 285
8GByte RAM
WIN 64 Pro, mATX board
MAX 6.1.9 Standalone (32bit)
CPU: 165.2
GPU Geometry 1: 248.4
GPU Geometry 2: 240.3
GPU Pixel Shaders: 249.5
MAX 7.0 (32bit, couldn´t build standalone because "cg.dll missing"?)
CPU: 193.4
GPU Geometry 1: 326.7
GPU Geometry 2: 244.5
GPU Pixel Shaders: 247.9
MAX 7.01 (32bit)
v2 Test
Extreme(!)drop in Geometry1 (?)
CPU: 196.1
GPU Geometry 1: 189
GPU Geometry 2: 248.9
GPU Pixel Shaders: 247.8
Bene
It's crazy how much better AMD graphics cards seem to perform on the high end. Good to know!
I tested a lot of machines, I will post them in separate posts for clarity.
First off: hackintosh -> so windows and mac sequencially
i7 920 2.7Ghz
18Gb RAM
GeForce GTX 660
------------------------
MAX 6.1.9 32bit windows 7 64bit
CPU: 107.5
GPU Geometry 1: 8.2
GPU Geometry 2: 39.5
GPU Pixel Shaders: 162.8
------------------------
MAX 6.1.9 64bit windows 7 64bit
CPU: 120.7
GPU Geometry 1: 8.0
GPU Geometry 2: 39.7
GPU Pixel Shaders: 163.6
------------------------
MAX 7 32 bit windows 7 64bit
CPU: 120.4
GPU Geometry 1: 7.9
GPU Geometry 2: 39.7
GPU Pixel Shaders: 197.6
-------------------------
MAX 7 64 bit windows 7 64bit
CPU: 132.5
GPU Geometry 1: 8.5
GPU Geometry 2: 39.7
GPU Pixel Shaders: 194.8
==================
MAX 6.1.9 - MAC 10.8.5 (performance is better with 10.9 and 10.10 I think)
CPU: 75.0 -> 97.6 second run
GPU Geometry 1: 8.4
GPU Geometry 2: 40.6
GPU Pixel Shaders: 112.0
----------------------------
MAX 7 - - MAC 10.8.5 (performance is better with 10.9 and 10.10 I think)
CPU: 103.7 - second run
GPU Geometry 1: 5.9
GPU Geometry 2: 38.8
GPU Pixel Shaders: 134.8
In the Apple store I tested with a MAX 6.1 stand alone - with max 7 it should be better as we saw!
---------
5k imac 3.5hgz i5 8gb ram amd radeon r9 m290x 2gbvram
109cpu
Geom 1 29.4
Geom2 200.9
Shader 210.3
----
13'' Macbook air 1.4 i5 4ram intel 5000
Cpu 70.1
21.7
96.7
41.8
-----------
13'' Macbook Pro 2,5 i5 4GB intel 4000
CPU: 73.8
GPU Geometry 1: 13.0
GPU Geometry 2: 80.8
GPU Pixel Shaders: 30.3
------------
MacBook Pro (Retina, 15 inch, medio 2014)
2,2 GHz Intel Core i7
16 GB 1600 MHz DDR3
Intel Iris Pro 1536 MB
"Copy these values:
CPU: 106.6
GPU Geometry 1: 28.0
GPU Geometry 2: 114.1
GPU Pixel Shaders: 67.3
----------------
2.3 i7 (late 2013 model)
16GB geforce GT 750M 2GB
CPU: 107.7
GPU Geometry 1: 6.2
GPU Geometry 2: 26.1
GPU Pixel Shaders: 92.2
I discovered a strange bug in Mac OS X. On laptops with two graphics 'cards' (intel and NVIDIA). There is the option in system prefs to turn on automatic switching on or off. Well turns out that performance is much worse if you leave that on. And not because it isn't switching to the NVIDIA (because with the app gfxCardStatus you can see it switch). Just a heads up for other Macbook Pro owners...
2.3 i7 (late 2013 model)
16GB geforce GT 750M 2GB
max 6.1.9 - Mac OS 10.10.1 - with auto graphics switching off
CPU: 108.1
GPU Geometry 1: 7.1
GPU Geometry 2: 26.3
GPU Pixel Shaders: 92.3
..........
max 6.1.9 - Mac OS 10.10.1 - with auto graphics switching on
CPU: 60.0
GPU Geometry 1: 6.7
GPU Geometry 2: 26.4
GPU Pixel Shaders: 59.9
----------
----------
MAX 7 - MAC OS 10.10.1 - graphics auto switching off
CPU: 174.9
GPU Geometry 1: 4.9
GPU Geometry 2: 26.5
GPU Pixel Shaders: 93.4
.........
MAX 7 - MAC OS 10.10.1 - graphics auto switching on
CPU: 60.0
GPU Geometry 1: 6.4
GPU Geometry 2: 26.7
GPU Pixel Shaders: 59.8
Kudos for raiding the Mac store :)
yeah no problem :) I come by it often enough and it is super insightful to see the comparisons!
Great to see also how better Max 7 performs on the CPU side. Thanks Cycling!
I might do it again with a Max 7 stand alone in the future. Too bad they don't have any laptops with graphics cards in the store. They only have the ones with integrated graphics. But of course it can be insightful in the future how they perform.
We really should make a website with the results or something. Maybe update the test with a few extra tests?
But I am already so happy that we did this run. Thanks Pedro Santos!
seem is best on Max 6.19
Max 7
GPU Geometry 1: 3.9
GPU Geometry 2: 27.9
GPU Pixel Shaders: 59.9
before max 6.19
MacBook Pro
Retina, 15-inch, Late 2013
2,3 GHz Intel Core i7
16 GB 1600 MHz DDR3
NVIDIA GeForce GT 750M 2048 MB
CPU: 104.9
GPU Geometry 1: 6.4
GPU Geometry 2: 26.9
GPU Pixel Shaders: 99.3
@Luxi you have the same model laptop as I. Strange. Did you have the graphics switching off in the system preference > energy?
The CPU part of the Max 7 is not showing, though...
What version of the OS are you on?
MacBook Pro
Retina, 15-inch, Late 2013
2,3 GHz Intel Core i7
16 GB 1600 MHz DDR3
NVIDIA GeForce GT 750M 2048 MB
Yosemite 10.10.1
Max 7
CPU: 168.5
GPU Geometry 1: 4.9
GPU Geometry 2: 27.6
GPU Pixel Shaders: 99.3
=======================
Max 6,19 same mbp, Maverick SO
CPU: 104.9
GPU Geometry 1: 6.4
GPU Geometry 2: 26.9
GPU Pixel Shaders: 99.3
I try with Win 7 pro
AMD 1090 six core
Nvidia quadro 2000
GPU Geometry 2: 440
Windows 7,64 Bit
CPU: i7-2700k
Ram: 16 GB
ASUS-HD7970
Max 6.1.9 32bit
CPU: 125.1
GPU Geometry 1: 166.1
GPU Geometry 2: 163.3
GPU Pixel Shaders: 177.2
Max 7.0 32bit
CPU: 175.8
GPU Geometry 1: 226.9
GPU Geometry 2: 174.8
GPU Pixel Shaders: 244.5
Max 7.0 64bit
CPU: 191.1
GPU Geometry 1: 226.6
GPU Geometry 2: 173.2
GPU Pixel Shaders: 242.5
I discovered something odd. I ran the test 5 times in a row.
Same computer. same settings
Max osx 10.9.5 retina 15"
Processor 2.8 Ghz i7
Ram: 16G
Here are the results
1.
CPU: 169.5
GPU Geometry 1: 6.4
GPU Geometry 2: 27.0
GPU Pixel Shaders: 86.3
2.
CPU: 187.6
GPU Geometry 1: 5.1
GPU Geometry 2: 25.4
GPU Pixel Shaders: 93.4
3.
CPU: 172.6
GPU Geometry 1: 3.9
GPU Geometry 2: 25.7
GPU Pixel Shaders: 86.6
4.
CPU: 169.6
GPU Geometry 1: 3.2
GPU Geometry 2: 27.5
GPU Pixel Shaders: 87.0
5.
CPU: 178.1
GPU Geometry 1: 2.5
GPU Geometry 2: 25.2
GPU Pixel Shaders: 86.7
MacBook Pro 15, early 2011
Os X 10.10
2,2 GHz Intel Core i7
SSD
AMD Radeon HD 6750M 1024 MB
16GB Ram
Max 6.1.9
CPU: 102.7
GPU Geometry 1: 19.7
GPU Geometry 2: 114.1
GPU Pixel Shaders: 45.7
Max 7.0
CPU: 141.6
GPU Geometry 1: 17.5
GPU Geometry 2: 115.0
GPU Pixel Shaders: 46.9
EDIT_____________
JITTER_BENCHMARK-V1.012.MAXPAT
_________________
A compilation of some results...
Laptop: Buy a lenovo with pro card for half the price of a MacBookPro and get 2x better geometry and 0,5x shading capabilities.
Desktop:
Buy a Win64 with AMD or Nvidia Card for half the price of a MacPro and get 2,5x overall speed.
Well, I knew it but always rejected it...
Sadly, time as come to let go my macs.
and take the PC way for realtime 3d.
Does anyone Know if a syphon equivalent is reliable on Win64?
MacBook Pro (Retina, 15-inch, Late 2013)
2 GHz Intel Core i7 – 16 GB 1600 MHz DDR3
Intel Iris Pro 1536 MB
OSX 10.10.1
Max 7.0
CPU: 155.1
GPU Geometry 1: 32.4
GPU Geometry 2: 151.5
GPU Pixel Shaders: 66.6
@spa: Pretty much my conclusion as well.
Thank you very much for you graph :-)
About the conclusion of MATTEO PENNESE in regards to laptops: is it so clear cut that a NVIDIA quadro is the best? The shader results are pretty low. Aren't the GPU calculations (slab, gen etc) of Jitter done on the shader?
Of course being able to display more polygons also counts for some thing. Eg. can I render more instances of geometry with this? For instance with multiple and phys?) I just wonder what will effect my frame rates most.
I'm looking into buying a Windows Laptop. But very much doubting between a Lenovo with a W2100M quadro and an other model with a GTX 860M.
Only on poly's does the W2100M have an advantage, on the rest the 860M seems to perform better?
I would love to hear some thoughts on this.
Shaders are (mostly) for processing of textures, so for example post-processing your rendered geometry or video streams. Shaders have no effect on geometry/scene rendering (see the GTX670 results, stellar shader fps, lousy geometry fps).
From the looks of all this, if I'd have to buy right now, I'd stay away from Nvidia and go for AMD. Too bad we don't have much mobile AMD results but the odds are they 're better across the board.
DTR: your right, but for a laptop I also have the consideration of wanting to work with the Kinect2. The Kinect2 needs preferably a USB3.0 chipset from intel.
Oh actually: I see that there are laptops with i7 and AMD R9 graphics... Let's see..
the nvidia card is showing poor results in the two gpu-geometry tests because those patches are using fixed function shading. most likely the newer cards are not optimized to perform fixed-function shading. simply connecting a gl.material object to the gl.multiple, and the 3 gl.gridshape objects in the gpu-geometry sub-patches gives similar results with the nvidia card.
Yep you're right,
Rob to the rescue !! :-)
Here are my new results
mbp 15"
OSX 10.9.4
Processor 2.8 GHz Intel Core i7
Memory 16 GB 1600 MHz DDR3
Graphics NVIDIA GeForce GT 750M 2048 MB
CPU: 170.4
GPU Geometry 1: 51.6
GPU Geometry 2: 198.1
GPU Pixel Shaders: 90.8
That's more like it! Attached is the adapted patch.
v1.012
i5 4690k 3.5GHz (slight OC)
GA-Z87X-OC mobo
EVGA (Nvidia) GTX670 2GB FTW gfx card
8GB RAM 1600Mhz
Win 7 pro 64 bit
max 6.1.9 first run
CPU: 126.0
GPU Geometry 1: 242.1
GPU Geometry 2: 236.2
GPU Pixel Shaders: 245.7
max 6.1.9 subsequent runs
CPU: 164.6
GPU Geometry 1: 175.3
GPU Geometry 2: 247.1
GPU Pixel Shaders: 246.4
max 7 first run
CPU: 197.2
GPU Geometry 1: 277.9
GPU Geometry 2: 307.8
GPU Pixel Shaders: 331.3
max 7 subsequent runs
CPU: 204.7
GPU Geometry 1: 243.1
GPU Geometry 2: 301.7
GPU Pixel Shaders: 331.5
I stated first and subsequent run results because there are weird discrepancies. Getting faster on 1 element and slower on the other...
Max 7 clearly improves performance over 6. Both tested in 32bit mode.
Hello Phiol,
I'm about to buy my new computer....
Is your Macbookpro retina 15,4" the model updated in July this year?
Cheers
Hubert
yep. I bought it on Oct 20th 2014
hope this helps
It's funny, I was just going to post something related to what Rob Ramirez wrote (welcome to the thread, Rob!).
When I wrote the original patch, I noticed discrepancies between using the fixed-function OpenGL pipeline and the current programmable pipeline (using shaders). For instance, in the GPU Geometry Test 1 (using jit.gl.multiple), with my AMD Radeon HD 4870, the results were something like:
fixed-function pipeline (plain OpenGL material): 100 fps
programmable pipeline (using shaders - jit.gl.material): 68 fps
So I thought that the standard OpenGL implementation would always be quicker.
As the tendency has been for a while to "deprecate the fixed-function pipeline (here's a nice article about the subject), probably more recent graphics cards' drivers don't optimise for it.
Thinking about this, I was going to suggest to build a complex OpenGL scene with objects using jit.gl.material, render-to-texture, depth buffers and post-processing effects (maybe Max7 shadows and other render passes).
Anyway, here's another nice lesson from this thread: USE JIT.GL.MATERIAL! It's probably faster! Who would have thought?
By the way, if that is the case, shouldn't Jitter's default behaviour be NOT to use the fixed-function pipeline? Or, at least, there could be a simple option similar to the OpenGL Readback mode present in the OpenGL Status menu? Just a thought...
DTR: nice numbers! Finally!
Thanks to MRMAARTEN's findings and visits to the Apple Store ;-) , thanks to SPA's results' compilation and thanks to everyone's contributions and for making this thread useful.
Btw, for direct comparison we're gonna need some of the AMD graphics owners to run the new version too. The outcome may be that the one shading method is better for Nvidia gfx and the other for AMD. Will be useful to be aware of this. Same for Intel integrated gfx.
Rob Ramirez - thanks for explaining that to us - i'd been a bit worried id just bought a new computer that was going to bottleneck due to an inexplicably crap graphics card - greatly appreciated.
hello dear tester...
is this computer hackintoshable lenovo-W520 2,2 quadro4000???
cheers
Hubert
Hackintosh OSX.9.5 i7 2700k AMD 6870 1 GB
Jitter Benchmark 1.012
Max 7.0.0
CPU: 174.6
GPU Geometry 1: 31.1
GPU Geometry 2: 185.0
GPU Pixel Shaders: 175.7
Max Runtime 6.0.8
CPU: 134.0
GPU Geometry 1: 30.4
GPU Geometry 2: 105.5
GPU Pixel Shaders: 172.7
@Generalh: this is really not the place to ask details about hackintoshing, try here : http://www.tonymacx86.com
@dtr I've already been there but man this is a mess with laptop stuff...So if somebody has already work it out it would have been easier... But yes tonymacx86 is the right place to start....
Thank you anyway
Mac Pro 6-core 3.33 Ghz 12GB RAM OSX 10.8.5 GeForce GTX680 2048MB
Benchmark v. 1.012
Max 7.0.0
CPU: 147.2
GPU Geometry 1: 54.2
GPU Geometry 2: 193.9
GPU Pixel Shaders: 182.9
Max 6.1.9
CPU: 109.2
GPU Geometry 1: 51.4
GPU Geometry 2: 143.1
GPU Pixel Shaders: 127.7
For comparison…
Benchmark v. 1.011
Max 7.0.0
CPU: 147.9
GPU Geometry 1: 10.8
GPU Geometry 2: 54.8
GPU Pixel Shaders: 181.7
Max 6.1.9
CPU: 108.8
GPU Geometry 1: 10.9
GPU Geometry 2: 55.6
GPU Pixel Shaders: 129.7
@Jesse: Interesting, your Mac system is kinda in the same league as my Win PC. Your GTX680 gfx card should be a bit faster than my GTX670 yet my system is faster by a large margin. This might point out OSX vs Windows performance differences. Would you happen to have a Bootcamp/Windows install on it for direct comparison?
Though I wonder how meaningful the results are when they're in the 200+fps range. I'll try to make a new/adapted test soon, with multi-display rendering etc to stress those high end systems more and see what comes out then. Got a bunch of gigs coming up again so no guaranteed ETA though.
No bootcamp install here but would be glad to run multi-screen tests. I agree the results are troubling regarding Mac/PC performance differences, it's likely a driver optimization issue.
Gosh golly what a difference! Also between win and mac a bit...
i7 920 2.7Ghz
18Gb RAM
NVIDIA GeForce GTX 660 2048 MB
Hackintosh 10.8.5
MAX7
CPU: 99.2
GPU Geometry 1: 34.5
GPU Geometry 2: 150.7
GPU Pixel Shaders: 131.3
MAX6.1
CPU: 86.0
GPU Geometry 1: 30.7
GPU Geometry 2: 101.4
GPU Pixel Shaders: 109.6
==================
Windows 7
Max7
CPU: 152.0
GPU Geometry 1: 175.1
GPU Geometry 2: 196.9
GPU Pixel Shaders: 199.3
Max6.1
CPU: 120.4
GPU Geometry 1: 161.9
GPU Geometry 2: 158.9
GPU Pixel Shaders: 165.8
MacBook Pro (Retina, 15-inch, Late 2013)
2,3 GHz Intel Core i7
16 GB 1600 MHz DDR3
NVIDIA GeForce GT 750M 2048 MB
MAX7 10.10.1
CPU: 168.5
GPU Geometry 1: 50.7
GPU Geometry 2: 176.0
GPU Pixel Shaders: 96.0
-----
MAX6.1.9 10.10.1
CPU: 109.9
GPU Geometry 1: 44.1
GPU Geometry 2: 140.8
GPU Pixel Shaders: 94.4
...
Macbook pro 2014 retina 15-inch
2.5 GHz Intel Core i7
16 GB 1600 MHz DDR3
NVIDIA GeForce GT 750M 2048 MB
Max7 10.10.1
CPU: 164.2
GPU Geometry 1: 6.4
GPU Geometry 2: 26.9
GPU Pixel Shaders: 94.3
---
Max6 10.10.1
CPU: 170.6
GPU Geometry 1: 7.1
GPU Geometry 2: 24.8
GPU Pixel Shaders: 104.1
Hi Wei I think you used the old Benchmark.
The thread is getting a bit long and confusing: I made a new one for V2 of the benchmark. I gave a little intro and posted the old results and relevant discussion.
Let's post our new results there: https://cycling74.com/forums/v2-benchmarking-jitters-cpugpu-performance-on-your-computer-v2/
I have put the new benchmark patch on top of the new thread as well. Hope this helps.
Posted my results for the V2 in the V2 thread... here are my results for the V1.011 patch
Mac Pro 6,1 (nMP)
6 core Xeon @ 3.5 GHz
AMD FirePro D700 (2x)
Max 7 64 bit mode
CPU: 250.7
GPU Geometry 1: 25.4
GPU Geometry 2: 186.0
GPU Pixel Shaders: 218.0
Max 7 32 bit mode
CPU: 197.2
GPU Geometry 1: 24.7
GPU Geometry 2: 182.4
GPU Pixel Shaders: 213.2
still embarrassing...
I have a Win 8 bootcamp but don't have Max installed their yet - Will run that test next time I am booted into Windows
Macbook Pro 8,2
Intel Core i7 2.2 gHz
16 GB RAM
AMD Radeom HD 6750M
CPU: 172.0
GPU Geometry 1: 20.5
GPU Geometry 2: 117.0
GPU Pixel Shaders: 47.6
HP z230
Xeon E3 3.4 gHz
16 GB RAM
Nvidia Quadro K2000
Windows 8.1
Max 7.0.1 64bit
Benchmark v1:
CPU: 200.2
GPU Geometry 1: 88.0
GPU Geometry 2: 437.7
GPU Pixel Shaders: 112.8
Benchmark v2 1st run:
CPU: 195.5
GPU Geometry 1: 56.2
GPU Geometry 2: 319.6
GPU Pixel Shaders: 112.5
MacBook Air Early 2014
8 GB 1600 MHz DDR3 / 1,4 GHz Intel Core i5
Intel HD Graphics 5000 1536 MB
OSX Yosemite
Max 7.1
CPU: 101.8
GPU Geometry 1: 21.7
GPU Geometry 2: 98.6
GPU Pixel Shaders: 37.1
Intel Core i7 4790K @4.0Ghz
32Gb RAM
GeForce GTX 760
CPU: 301.9
GPU Geometry 1: 10.5
GPU Geometry 2: 49.2
GPU Pixel Shaders: 491.5
CPU: 316.3
GPU Geometry 1: 7.0
GPU Geometry 2: 49.2
GPU Pixel Shaders: 491.5
How can I tell that a given patch is using 'fixed function' shading? How does connecting [jit.gl.material] to [jit.gl.multiple] remedy the issue (i.e. how/why does it convert the patch to non-fixed function)?
Not only are my 'fixed function' benchmarks low - scores gradually drop in performance by close to 70% after 5 or 6 bench tests during the same Max session.. crawling as low as 2.7 on a 2 year old iMac... (using version 1.011 of the test patch.. the newest v1.012 much less so).
Rob Ramirez wrote:
the nvidia card is showing poor results in the two gpu-geometry tests because those patches are using fixed function shading. most likely the newer cards are not optimized to perform fixed-function shading. simply connecting a gl.material object to the gl.multiple, and the 3 gl.gridshape objects in the gpu-geometry sub-patches gives similar results with the nvidia card.
fixed function just means any object drawn without a shader / material applied. if you enable lighting and smooth shading on an object instead of simply attaching a jit.gl.material, the fixed function pipeline will be used.
Computer MSI GS60 2QE (laptop) with Max 7.3.4 64bit on Windows 10
Intel Core i7 4710HQ @2.5GHz
8Go RAM
GeForce GTX970M with 3Go VRAM
CPU: 159.3
GPU Geometry 1: 194.7
GPU Geometry 2: 136.6
GPU Pixel Shaders: 205.9
and with the integrated HD Graphics 4600
CPU: 153.7
GPU Geometry 1: 6.8
GPU Geometry 2: 116.5
GPU Pixel Shaders: 33.6
2011 MacBook Pro 17"
2.2 GHz Intel Core i7
16 GB 1333 MHz DDR3 RAM
AMD Radeon HD 6750M 1024 MB
CPU: 151.2
GPU Geometry 1: 18.8
GPU Geometry 2: 119.2
GPU Pixel Shaders: 47.1
Can someone explain these values?