Return-Path: Received: from mail-ob0-f175.google.com ([209.85.214.175] verified) by media-motion.tv (CommuniGate Pro SMTP 4.2.10) with ESMTP-TLS id 5423180 for AE-List@media-motion.tv; Wed, 02 Apr 2014 20:34:34 +0200 Received: by mail-ob0-f175.google.com with SMTP id uy5so742045obc.6 for ; Wed, 02 Apr 2014 11:35:59 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :content-type; bh=22UWrb0cXqcnSIANqIxl9/7a3QTW8cvjBjyKdcmM764=; b=VsftCfM3BlOCei9EgHbtXq8gh6gXP9OV2Q5LjIryxrF7XTLVugTdUTK0YprysESVPt +lTQgFI5TTeZsPYLkeEbKuizbP2Ok/vbbQP0G2hlG7K7geb5S2eyH50fOpJuLYONA6Ls TSCy7w3SFm6lhiq2LFVO6aMp8gAsj7e78gPjLdQ3B0SnI1Y6myk+KqC0s22pckZ81ERK qNCJNajAzW6BwUC/BOmI672ZJpv945rRiZRL2aLySoXfkeJEtSEBEM7v6yXiyvPdnz6Y t89xYm7kRH2+PXktyA0VZJ3Ib3ddYJhctKPWYKBbKX8VfwjKoXY7BRoSMAY+ZvGIDvhY cJng== MIME-Version: 1.0 X-Received: by 10.182.250.200 with SMTP id ze8mr1391194obc.72.1396463759297; Wed, 02 Apr 2014 11:35:59 -0700 (PDT) Received: by 10.76.173.134 with HTTP; Wed, 2 Apr 2014 11:35:59 -0700 (PDT) In-Reply-To: References: Date: Wed, 2 Apr 2014 14:35:59 -0400 Message-ID: Subject: Re: [AE] AfterBenchCC - A free total After Effects benchmarking suite for CPU/GPU/RAM From: rendernyc To: After Effects Mail List Content-Type: multipart/alternative; boundary=089e0160c6605091fb04f6138e72 --089e0160c6605091fb04f6138e72 Content-Type: text/plain; charset=ISO-8859-1 I've used chud tools and the processor pane to dsiable cores in the past but it seems theres a terminal way to do it more permanently https://developer.apple.com/library/mac/qa/qa1141/_index.html On Wed, Apr 2, 2014 at 1:00 PM, Teddy Gage wrote: > Byron, definitely let me know. This is exactly why I created this > project by the way - to catch and eliminate systems bottlenecks and > maximize performance on people's machines. For example, I discovered that, > at least according to these tests, that unless you are running Cineware or > Element 3D, you should turn OFF hyper-threading for a huge performance > boost in AE. This is done in the BIOS on PC machines, usually under CPU > performance settings. Unfortunately Apple has disabled this functionality > in OS versions past 10.6, which is just insane to me. However you can set > "reserved cores" in the AE prefs to be half your logical cores, equal to > your physical cores, and it should help. This way the maximum amount of RAM > is available to each core and should improve render performance. > -TG > > > On Wed, Apr 2, 2014 at 12:24 PM, Byron Nash wrote: > >> Yeah, something seems fishy. I got 88fps with the 680 on a PC. I may try >> to put the Titan in the PC next and see how it fairs. >> >> >> On Wed, Apr 2, 2014 at 11:59 AM, Byron Nash wrote: >> >>> Yeah, I'm getting 45fps in Cinebench. 1288cb for the CPU. I'll test my >>> PC with a 680 next and see how it compares. >>> >>> >>> On Wed, Apr 2, 2014 at 11:20 AM, Teddy Gage wrote: >>> >>>> Hmmm interesting... Have you run any other tests on the titan? My guess >>>> is that there is some kind of bottleneck on the motherboard, maybe the pci >>>> bandwidth of the older mac pro doesnt support all the titan throughput? Or >>>> could be that the pc drivers are simply much more optimized? >>>> >>>> What kind of numbers do you get from this test >>>> http://forums.creativecow.net/thread/2/1019120 >>>> >>>> Should be around 4-6 minutes i think >>>> >>>> Or from cinebench r15 in the openGL portion? I get around 110 - 120 FPS >>>> in the openGL test on that with the titan, depending on overclock >>>> >>>> >>>> On Wednesday, April 2, 2014, Byron Nash wrote: >>>> >>>>> I just ran the test on a MacPro(5,1) 12 core with 16GB RAM and a Titan >>>>> GPU. I only got 12min on the GPU test. I wonder why it's so much slower >>>>> than your PC with the Titan? Do you think the CPU speed or RAM is affecting >>>>> the performance in that GPU heavy comp? >>>>> >>>>> >>>>> On Tue, Apr 1, 2014 at 1:34 PM, Teddy Gage wrote: >>>>> >>>>> Greg, here are the results so far, will make it prettier once I get >>>>> more info. Please send me your scores to fstopdigital@gmail.com or on >>>>> twitter @teddyrocksteady >>>>> >>>>> http://goo.gl/JrNIXd >>>>> >>>>> Would love to see this run on a new mac pro, any takers? Should only >>>>> take about 20 minutes >>>>> >>>>> >>>>> On Mon, Mar 31, 2014 at 12:47 PM, Greg Balint < >>>>> delrazoraelist@gmail.com> wrote: >>>>> >>>>> Hey Teddy, >>>>> >>>>> Any way you could do something similar to cbscores.com with the >>>>> results? >>>>> >>>>> Maybe ask the 3d Fluff guys for a template of their page with the >>>>> submit functionality and such. It's been a great go-to when researching new >>>>> processors to see how they fair in the benchmark. >>>>> >>>>> ///Greg Balint >>>>> //Art Director / Motion Graphics Designer >>>>> /321.514.4839 >>>>> delRAZOR.com/ >>>>> >>>>> On Mar 31, 2014 11:46 AM, "Teddy Gage" wrote: >>>>> >>>>> New update - Here is a sample video of all the files that your machine >>>>> will render in a run of the 60 MB AfterBenchCC project >>>>> >>>>> https://vimeo.com/90489125 >>>>> >>>>> Again, the file is available at www.teddygage.com/AfterBenchCC/ >>>>> >>>>> It's not really recommended for systems older than 2 years, it may >>>>> take a long time to run, especially with older GPUs. However it is broken >>>>> into several 3-5 minute tests, so you can pick and choose which test you >>>>> want to run without demanding too much time or resources. I highly >>>>> recommend the fractal noise CPU test to get a clear idea of system >>>>> performance in a single run. >>>>> >>>>> Here are my current results, the rest of our machines will be added >>>>> soon. >>>>> >>>>> "IRONMAN" >>>>> 4.7 GHz hex core CPU, 64 GB RAM, 720 GB SSD RAID1, hyperthreading off >>>>> >>>>> MP OFF: 2:28 >>>>> 4K KEYFRAME: 3:12 >>>>> 4K ENCODING H264: 0:26 >>>>> Fractal Noise 4K all CPU: 3:55 >>>>> 4K Encoding MP4: 0:18 >>>>> RT & EnvMap GPU: 5:12 >>>>> Cineware: 8:51 >>>>> Element 3D: 5:02 >>>>> >>>>> >>>>> "The flash" >>>>> 4.3 GHZ quad cpu, 4770k, 16gb ram, 2x gtx 580, hyperthreading on >>>>> >>>>> MP OFF: 2:22 >>>>> 4K KEYFRAME: 3:25 >>>>> 4K ENCODING H264: 0:36 >>>>> Fractal Noise 4K all CPU: 7:11 >>>>> 4K Encoding MP4: 0:22 >>>>> RT & EnvMap GPU: 4:15 >>>>> Cineware: 8:42 >>>>> Element 3D: na >>>>> >>>>> >>>>> "JACKAL" >>>>> Late 2013 Macbook pro >>>>> 2.6 GHZ boost up to 3.8 GHz, quad core cpu, 16gb ram, gtx 720m, 1 TB >>>>> on board SSD, hyperthreading on >>>>> >>>>> MP OFF: 2:34 >>>>> 4K KEYFRAME: 3:36 >>>>> 4K ENCODING H264: 0:35 >>>>> Fractal Noise 4K all CPU: 7:01 >>>>> 4K Encoding MP4: 0:25 >>>>> RT & EnvMap GPU: 43:23 >>>>> Cineware: 9:13 >>>>> Element 3D: na >>>>> >>>>> Some really interesting numbers here, I'm going to create a >>>>> spreadsheet with some nice graphs and have a link on my site. Thanks for >>>>> checking it out! >>>>> -TG >>>>> >>>>> >>>> >>>> -- >>>> _____________________________ >>>> VFX & Motion Graphic Artist >>>> teddygage dot com >>>> >>>> >>> >> > > > -- > _____________________________ > VFX & Motion Graphic Artist > teddygage dot com > -- danny princz vimeo.com/rendernyc --089e0160c6605091fb04f6138e72 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable
I've used chud tools and the processor pane to dsiable= cores in the past but it seems theres a terminal way to do it more permane= ntly




On Wed, Apr 2, 2014 at 1:00 PM, Teddy Gage <teddygage@gmail.com> wrote:
=A0 =A0 Byron, definitely l= et me know. This is exactly why I created this project by the way - to catc= h and eliminate systems bottlenecks and maximize performance on people'= s machines. For example, I discovered that, at least according to these tes= ts, that unless you are running Cineware or Element 3D, you should turn OFF= hyper-threading for a huge performance boost in AE. This is done in the BI= OS on PC machines, usually under CPU performance settings. Unfortunately Ap= ple has disabled this functionality in OS versions past 10.6, which is just= insane to me. However you can set "reserved cores" in the AE pre= fs to be half your logical cores, equal to your physical cores, and it shou= ld help. This way the maximum amount of RAM is available to each core and s= hould improve render performance.
-TG


On Wed, Apr 2, 2014= at 12:24 PM, Byron Nash <byronnash@gmail.com> wrote:
Yeah, something seems fishy= . I got 88fps with the 680 on a PC. I may try to put the Titan in the PC ne= xt and see how it fairs.


On = Wed, Apr 2, 2014 at 11:59 AM, Byron Nash <byronnash@gmail.com> wrote:
Yeah, I'm get= ting 45fps in Cinebench. 1288cb for the CPU. I'll test my PC with a 680= next and see how it compares.


On Wed, Apr 2= , 2014 at 11:20 AM, Teddy Gage <teddygage@gmail.com> wrote= :
Hmmm interesting... Have you run any other t= ests on the titan? My guess is that there is some kind of bottleneck on the= motherboard, maybe the pci bandwidth of the older mac pro doesnt support a= ll the titan throughput? Or could be that the pc drivers are simply much mo= re optimized?

What kind of numbers do you get from this test=A0

Should be around 4-6 minutes i think

Or from cinebench r15 in the openGL portion? I get arou= nd 110 - 120 FPS in the openGL test on that with the titan, depending on ov= erclock


On Wednesday, April 2, 2014, Byron Nash <byronnash@gmail.com> wrote:
I just ran the t= est on a MacPro(5,1) 12 core with 16GB RAM and a Titan GPU. I only got 12mi= n on the GPU test. I wonder why it's so much slower than your PC with t= he Titan? Do you think the CPU speed or RAM is affecting the performance in= that GPU heavy comp?


On Tue, Apr 1, 2014 at 1:34 PM, Teddy Gage <teddygage@gmail.com> wrote:
Greg, here are the results so far, will make it prettier o= nce I get more info. Please send me your scores to fstopdigital@gmail.co= m or on twitter @teddyrocksteady

http://goo.gl/JrNIXd=A0

Would love to see this run on a new mac pro, any takers= ? Should only take about 20 minutes

On Mon, Mar 31, 2014 at 12:47 PM, Greg Balint <delr= azoraelist@gmail.com> wrote:

Hey Teddy,

Any way you could do something similar to cbscores.com with the results?

Maybe ask the 3d Fluff guys for a template of their page wit= h the submit functionality and such. It's been a great go-to when resea= rching new processors to see how they fair in the benchmark.

///Greg Balint
//Art Director / Motion Graphics Designer
/321.514.4839
delRAZOR.com/
=A0=A0=A0

On Mar 31, 2014 11:46 AM, "Teddy Gage" <<= a>teddygage@gmail.com> wrote:
New update - Here is a sample video of all the files that = your machine will render in a run of the 60 MB AfterBenchCC project

Again, the file is available at www.teddygage.com/After= BenchCC/

It's not really recommended for s= ystems older than 2 years, it may take a long time to run, especially with = older GPUs. However it is broken into several 3-5 minute tests, so you can = pick and choose which test you want to run without demanding too much time = or resources. I highly recommend the fractal noise CPU test to get a clear = idea of system performance in a single run.=A0

Here are my current results, the rest of our machines w= ill be added soon.

"IRONMAN"
4.7 GHz hex core CPU, 64 GB RAM, 720 GB SSD RAID1, hyperthreading off

MP OFF: 2:28
4K KEYFRAME: 3:12
4K ENC= ODING H264: 0:26
Fractal Noise 4K all CPU: 3:55
4K Enco= ding MP4: 0:18
RT & EnvMap GPU: 5:12
Cineware: 8:51
Element 3= D: 5:02


"The flash"
=
4.3 GHZ quad cpu, 4770k, 16gb ram, 2x gtx 580, hyperthreading on

MP OFF: 2:22
4K KEYFRAME: 3:25
4K E= NCODING H264: 0:36
Fractal Noise 4K all CPU: 7:11
4K En= coding MP4: 0:22
RT & EnvMap GPU: 4:15
Cineware: 8:= 42
Element 3D: na


"JACKAL&qu= ot;=A0
Late 2013 Macbook pro
=A02.6 GHZ boost up to 3.8= GHz, quad core cpu, 16gb ram, =A0gtx 720m, 1 TB on board SSD, hyperthreadi= ng on

MP OFF: 2:34
4K KEYFRAME: 3:36
4K E= NCODING H264: 0:35
Fractal Noise 4K all CPU: 7:01
4K En= coding MP4: 0:25
RT & EnvMap GPU: 43:23
Cineware: 9= :13
Element 3D: na=A0

Some real= ly interesting numbers here, I'm going to create a spreadsheet with som= e nice graphs and have a link on my site. Thanks for checking it out!
-TG



--
_____________________________=
VFX & M= otion Graphic Artist
teddygage dot com






--
=
_____________________________=
VFX & Motion Graphic Artist
teddygage dot com



--
= --089e0160c6605091fb04f6138e72--