OCUS Benchmark - Video Cards (Quadro, Firepro)

[ptc-2889084]

We are evaluating new workstations and video cards to run Creo Parametric 2.0 M100. I had some unexpected results and would appreciate some comments.

We use Dell workstations. We evaluated two T3610 workstations running Windows 7 64 bit:

1. Xeon E5-1650v2 6 core 3.5 GHz 12MB cache 64 GB (machine #1)

2. Xeon E5-1620v2 4 core, 3.7 GHz 10 MB cache 64 GB (machine #2).

Although the machines were connected to the network (for licenses), all software and data files were local on the C: drive. The machine were procured a few months ago.

The video cards:

1. Nvidia Quadro K2000 (vc #1)

2. AMD Firepro W5000 (vc #2)

3. Nvidia Quadro K4000 (vc #3)

Running the OCUS benchmark (64 bit) (Thanks, Olaf) using the latest, not PTC certified, drivers yielded the following results (times in seconds):
Machine #1: VC #1 VC #2 VC #3
This thread is inactive and closed by the PTC Community Management Team. If you would like to provide a reply and re-open this thread, please notify the moderator and reference the thread. You may also use "Start a topic" button to ask a new question. Please be sure to include what version of the PTC product you are using so another community member knowledgeable about your version may be able to assist.

[StephenW]

A couple years ago I had trouble with the video card performance until my IT guy figured out he had to upgrade the chipset drivers (motherboard drivers) so it could use the better video cards.

If your computer are older, you might try that.

[iceblink1]

It is never easy to explain benchmark results (let alone predict them),
so it's always good to benchmark.
I'll try and explain some of this below.

[bgruman]

Well said Patrick!
I agree with everything Patrick pointed out and might only add one more
point.
In general, (I know there is some crossover so bear with me a bit...) Xeon
Processors require "server class" motherboards, and server class
motherboards are just plain slow. (on purpose... They are built for
reliability, not speed.)
Every Xeon based system I have ever built and tested is just slow compared
to an Overclocked i7 based system.
I do custom build Xeon based workstations for special applications like
heavy FEA, or multi-threaded rendering work. But in my opinion (and I have
many satisfied customers who agree) for the average everyday Pro/E user,
there is nothing faster than an Overclocked i7 based system.

Thanks
Bernie

Bernie Gruman
Owner / Designer / Builder
www.GrumanCreations.com



On Wed, May 14, 2014 at 2:47 AM, Patrick Asselman <-<br/>> wrote:

> It is never easy to explain benchmark results (let alone predict them),
> so it's always good to benchmark.
> I'll try and explain some of this below.
>
>

[jcrockett-2]

Bernie is absolutely correct. A Core i7 with the fastest clock speed will run ProE/Creo the fastest. However, overclocking is not really an option at manylarge companies. I push ProE pretty hard on my HP z420 workstation:

Intel Xeon E5-1620 3.6Ghz 10MB cache 4 cores

16GB DDR3-1600 ECC (4x4GB)

128GB SSD 1st drive, 500GB 10k SATA 2nd drive

NVIDIA Quadro K2000 2GB

...and I'm pretty satisfied with performance, with a reasonable price.

[EricVidra]

We are benchmarking (OCUSB6) Dell Workstations including a T1700. I find the same results with much higher overall score, nearly double the submitted result of #10 Dell T1700 @1153 we are reporting 2146. (64-bit version)

Looking closer we find the last 2 tests PDF and DXF file create much higher scores in 550 range. This is obviously influencing the overall score. Any thoughts?

Bob what scores did you have for the PDF and DXF file create?

Thanks

Eric Vidra

“I am employed by Toyota but I am not speaking on behalf of the Company. These are my personal opinions only.”

[ocorten-2]

Are you sure you use only the configuration of the benchmark?Remove all your own configuration files and start creo via a dos prompt in the ocusb6 folder (call the full path to parametric.exe)Hope this helps.


Eric Vidra --- [proesys] - RE: OCUS Benchmark - Video Cards (Quadro, Firepro) ---
From:"Eric Vidra" <->To" <->Date:Wed, Jun 4, 2014 23:05Subject[proesys] - RE: OCUS Benchmark - Video Cards (Quadro, Firepro)

We are benchmarking (OCUSB6) Dell Workstations including a T1700. I find the same results with much higher overall score, nearly double the submitted result of #10 Dell T1700 @1153 we are reporting 2146. (64-bit version)
Looking closer we find the last 2 tests PDF and DXF file create much higher scores in 550 range. This is obviously influencing the overall score. Any thoughts?
Bob what scores did you have for the PDF and DXF file create?
Thanks
Eric Vidra


“I am employed by Toyota but I am not speaking on behalf of the Company. These are my personal opinions only.”
Site Links: View post online View mailing list online Start new thread via email Unsubscribe from this mailing list Manage your subscription


Use of this email content is governed by the terms of service at:

[EricVidra]

Thanks, Yes, clean OCUSB6 folder, with no configs other than yours. I was using the alternative method described in readme file with creo shortcut start script edited to start in OCUS benchmark folder.




[cid:image002.jpg@01CF8000.760CA810]

[ocorten-2]

Hi Eric,

How much RAM does your system have?
If it's 8Gb or less you might experience problems with memory becoming full.
Then you would actually see the last tests become extremely slow, like you see now.

Which build of Creo 2.0 are you using?
Maybe it's a specific build issue.


Kind regards,

Olaf Corten




Olaf Corten | CAD/PLM Manager

Besi Netherlands B.V. | Ratio 6| 6921RW Duiven| The Netherlands
T: +31 26 3196215 | M: +31 644548554
- | www.besi.com



From: Eric Vidra <eric_vidra@toyota.com>
To: Olaf Corten <->
Cc: "-" <->
Date: 04-06-2014 23:24
Subject: RE: Re:[proesys] - RE: OCUS Benchmark - Video Cards (Quadro, Firepro)



Thanks, Yes, clean OCUSB6 folder, with no configs other than yours. I was using the alternative method described in readme file with creo shortcut start script edited to start in OCUS benchmark folder.

[ptc-2889084]

Eric, PDF 91 sec, DXF 74 sec

Based on Bernie Gruman's comments, I tested a Dell M6800 laptop which has an i7 processor with 32GB main memory, a Quadro K4100M graphics card and a 500GB solid state hybrid drive. Compared to the T3610 overall score of 1628, the M6800 came in at 978 seconds with an external monitor set to full resolution. That is about 60% less. A dell T1700 desktop can be configured with an i7; I am hoping to be able to test one.

A word of caution. I had some of our engineers replicate everyday design tasks. In general, the times on the M6800 were typically only 10-20% faster. One set of tests was faster on the T3610. The insight gained from the OCUS benchmark needs to be combined with your design practices.

Bob Monat

[EricVidra]

Olaf,

3 test machines system RAM 16GB, 32GB, 64GB.
CREO 2.0 M110

Each machine was very slow for last 2 tests. Yes Jered also suggested this could be build specific. I will run again with another build to verify.

Interestingly during my preparations and setup for OCUS, I did remote connection to machine and ran initial test, of course the graphic scores were way off, but the PDF DXF are very good:

(T1700 16GB remote)
04-Jun-14 03:56:04 = 62 seconds = 1 PDF file creations (CP+DI)
04-Jun-14 03:56:55 = 51 seconds = 1 DXF File creations (CP+DI)

(T1700 16GB direct login)
Later during the real testing logged in at machine I get very poor scores for same tests. All others test good:

04-Jun-14 07:17:03 = 558 seconds = 1 PDF file creations (CP+DI)
04-Jun-14 07:26:11 = 548 seconds = 1 DXF File creations (CP+DI)

Corrected total score would be in the 1150 range….and compares well with submitted results.



Thanks
Eric





“I am employed by Toyota but I am not speaking on behalf of the Company. These are my personal opinions only.”

[EricVidra]

Bob,

Thanks for the numbers. Yes we are also including some typical Engineer design tasks in our review. It was in these test we found surprisingly good numbers for the T1700 against higher spec machines, and turned to the OCUS for more understanding. I will share summary when complete.

Eric Vidra

[EricVidra]

Appears to be build related issue with M110. Here are latest numbers with M070.

T1700 very strong showing for a Xeon with no overclocking. Still wonder if we could get bit more performance with E5-1620 v2 3.7GHz for nearly same price in a T3610? But Bob Monat Machine #2 VC#3 did fair better with this CPU - on paper higher clock and much larger memory capacity.

[ocorten-2]

Hi all,

I see exactly the same results on my Dell T1700 with Creo Parametric 2.0.
M110 shows about 10 times slower PDF and DXF creation times compared to M100 (OCUS Benchmark v6 64bit - tests 36 and 37)
http://www.proesite.com/cgi-bin/ocusb6.cgi?v=All&srt=total&cmp=0&sub=1&srch=benchmark+t1700&sfrm=0&lid=20&64bit=1


(and I also see it on my Dell M6500 with the normal benchmark)


The problem goes away when using the config setting: windows_browser_type set to mozilla_based_browser.
The results for DXF and PDF creation are then back to normal (practically the same as M100).
This option was not included in the OCUS Benchmark config.pro. The default for this option is ie_in_process for M100 as well as M110.
You can include it in the config.pro manually, but I have also updated the benchmark config.pro (download version 6.0.1)
Note: I'm running Creo Parametric unlinked to Windchill ...

I have created a medium severity case on PTC.com (#12026663).


Kind regards,

Olaf Corten




Olaf Corten | CAD/PLM Manager

Besi Netherlands B.V. | Ratio 6| 6921RW Duiven| The Netherlands
T: +31 26 3196215 | M: +31 644548554
- | www.besi.com



From: Eric Vidra <->
To: "Monat, Robert A" <->, Olaf Corten <->
Cc: "-" <->
Date: 05-06-2014 22:02
Subject: [proesys] - RE: OCUS Benchmark - Video Cards (Quadro, Firepro)



Appears to be build related issue with M110. Here are latest numbers with M070.

T1700 very strong showing for a Xeon with no overclocking. Still wonder if we could get bit more performance with E5-1620 v2 3.7GHz for nearly same price in a T3610? But Bob Monat Machine #2 VC#3 did fair better with this CPU - on paper higher clock and much larger memory capacity.

http://cpuboss.com/cpus/Intel-Xeon-E5-1620-v2-vs-Intel-Xeon-E3-1270-v3



Thanks to all for all the helpful replies!
Eric



Dell Workstations Win7 64-bit, all with same SSD drive configurations (3) 840 EVO.
CREO 2.0 M070

T5610 #1
CPU: Dual Xeon 8 core E5-2687W v2 3.40 GHz Max Turbo 4.0
RAM: 64 GB
VIDEO: K5000 (4GB)

Total: 1769
CPU: 1345
Graphics: 402
Disk: 384

T5610 #2
CPU: Single Xeon 4 core E5-2637 v2 3.50 GHz Max Turbo 3.8
RAM: 32 GB
VIDEO: K5000 (4GB)

Total: 1725
CPU: 1228
Graphics: 478
Disk: 264

T1700 #3
CPU: Single Xeon 4 core E3-1270 v3 3.50 GHz Max Turbo 3.9
RAM: 16 GB
VIDEO: K4000 (3GB)

Total: 1107
CPU: 789
Graphics: 305
Disk: 168




“I am employed by Toyota but I am not speaking on behalf of the Company. These are my personal opinions only.”