Skip to main content
S
23-Emerald III
StephenW23-Emerald III
23-Emerald III
May 31, 2012
Question

Quadro 5000 VS Quadro FX3800

  • May 31, 2012
  • 7 replies
  • 1851 views
I have been benchmarking (OCUS) a new setup that I thought would give some improvement but when I replaced the video card FX3800 with a 5000.

Original System setup:
HP Z400
XEON 3.2 ghz
8 gig ram
Win 7 enterprise 64 bit
Quadro FX3800

I got pretty good results on the benchmark beating some other computers we currently use by 10% or so.

The IT guy had more ram (12 gig) and a Quadro 5000 available so we switched them out and re-ran the benchmark 3 times. I know the ram would help particularly since the benchmark doesn't get close to running out.

I am now getting about double the time (or twice the time to complete) for the benchmark. Double the graphics portion, 50% increase on the CPU portion, and break even on the disk portion.

Between the 1st and 2nd benchmark we tested the ram and it showed good.

Between the 2nd and 3rd benchmark we re-seated the ram and updated to the lastest and greatest video driver.

Any thoughts on what tweak I am missing?

In the long run, we will probably go with a z420 (with 3.6 processor) and 12 to 16 gig ram but I am really trying to justify the $500-$1000 extra on the video card. Everything else is a no brainer.





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.

7 replies

S
23-Emerald III
StephenW23-Emerald IIIAuthor
23-Emerald III
May 31, 2012
Oops.
Forgot, we are testing an solid state hard drive as primary. It is consistent in both test and according to the benchmark, it has performed equally (within a couple %)

Steve
S
23-Emerald III
StephenW23-Emerald IIIAuthor
23-Emerald III
May 31, 2012
It’s only one box. Just switching components and comparing against itself.

I am comparing to existing workstations but with old benchmark data that I saved. I'm running the same benchmark I downloaded a while back, so I didn't change benchmarks.


Original test was with "whatever" driver that was installed when it had the 3800 in it and then we switch video cards with no driver change and got bad performance. The updated to lastest drive still with bad performance.

Resolution might have changed with the card change, I will check that.

I don't think the RAM did anything for the benchmark, good or bad, I really suspect all the problem is with the video card settings.

This is all 64bit.

Steve


B
bgruman1-Visitor
1-Visitor
May 31, 2012
Steve,

You're seeing exactly what I would expect to see based on my past testing
with OCUS.

I would have to do some digging to give you some exact numbers and machine
specs, but in general here is what I have found...

Quadro FX3800 (Fastest testing card I have every used)
Quadro 4000 (Slightly slower than the FX3800, but still quite respectable)
Quadro 5000 (Even slower yet from the 4000) Unless you're pulling huge
assemblies that require this amount of on board video memory, these cards
are not worth the extra money in my opinion.

As you said in a different e-mail... When you look at the specs and
performance results on the Nvidia web site, you would think that the
Quadro 5000 would really make Pro/E sing, but in the specific testing of
Pro/E with the OCUS benchmark it doesn't work out that way. In short,
The older generation Quardo FX engines were testing faster than the new
ones. However, the FX3800 card (from what I can tell) is no longer in
production, and is getting very hard to find. (Not sure where HP is
getting them from) So, the new generation Quadro 4000 card has become my
card of choice for Pro/E Workstations. It is slower than the old FX
cards, but only slightly slower, and hopefully as the driver software
matures, NVidia will be able to pull even more speed out of these cards.

Another Note: Version of Pro/E vs Quadro cards.
There is something big going on with Creo1. (Which in my opinion is why
PTC is dropping Creo1 support so quickly and encouraging everyone to go to
Creo2) When using the latest OCUS v6 test and comparing speed results
between WF5, Creo1, and Creo2. Creo1 is super slow on the older hardware.
Check the OCUS scores and look at my GC-i7 machine... Creo2=1841, and
WF5=2090, but Creo1=2252! Then look at my GC-i7K scores and we see
something I would expect to see. Creo2=1132, Creo1=1295, and WF5=1463.
Basically, Creo1 does not play well with the older Quadro FX graphics
engine.

Hard drives: I have done lots and lots of testing with SSD's, vs 10K, vs
7200RPM drives. In the end, what I've found is that Pro/E really doesn't
care. so, from a Pro/E standpoint, SSD's are fastest, 10K drives are
slightly slower (within a few seconds), and 7200RPM drives as slightly
slower again. (within 15 seconds or so) Now, from an OS, and
Application "launch and save" point of view... SSD's are king, 10K's are
pretty good, and 7200's are OK, but feel really slow when you're used to
the speeds of SSD's.

Now another note aside from all above... The real "Bang for your buck" in
Pro/E performance comes from processor and RAM speed. (as shown in the
GC-i7K machine on proesite.com) But that's a whole other subject... Smiley Wink

Good Luck with your testing!
...and of course I have to throw in a plug...
If your interested in considering a GC machine, I would be happy to send
you a quote. Also, I could build you a GC-i7K machine for side-by-side
testing against the HP you're currently considering.

Take Care
Bernie

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



On Thu, May 31, 2012 at 10:25 AM, Williams, Steve C <
-> wrote:

> Oops. ****
>
> Forgot, we are testing an solid state hard drive as primary. It is
> consistent in both test and according to the benchmark, it has performed
> equally (within a couple %)****
>
> ** **
>
> Steve****
>
> ** **
>
> *From:* Glen R Wisham [
P
PaulSagar12-Amethyst
12-Amethyst
June 1, 2012
Steve
Can you confirm the version of Creo Parametric you are testing? We have done extensive graphics work in Creo 2.0 that will take more advantage of the graphics card. In earlier releases you would not see a significant difference in performance between cards. In Creo 2.0 you should be seeing jumps in performance as you switch to faster cards.

What you are seeing does seem as if you are testing a pre-Creo 2.0 release.

Please let me know if this is the case or not

Thanks in advance

Paul Sagar
Director of Product Management
PTC
S
23-Emerald III
StephenW23-Emerald IIIAuthor
23-Emerald III
June 1, 2012
Thanks Bernie. Your participation in the user group is REALLY helpful. We are running WF4 with no immediate plans to upgrade, so I won't be able to justify the Quadro 5000. We will probably go with the 4000 or FX3800 if available and I have already made my case for the fastest processor and ram speed.

Years ago when I worked for a smaller company, we were able to order computers through a custom builder and they were tuned up really nicely and you guys do so much better on personalized setups that WAY outperform the big companies like HP and Dell. But alas, being a big corporation, we are stuck with the mentality that we have to buy from another big corporation.

Thanks again
Steve
S
23-Emerald III
StephenW23-Emerald IIIAuthor
23-Emerald III
June 1, 2012
Paul,
We are running WF4 with no immediate plans on upgrading.

Thanks
Steve
S
23-Emerald III
StephenW23-Emerald IIIAuthor
23-Emerald III
June 1, 2012
Olaf,
We are running WF4, so I’m running OCUS 5 64 bit.

I had very specifically searched your benchmark yesterday looking at the 3 or so 5000’s on there. They did get decent performance.

I was getting very good results on this same machine with the FX3800 but with the 5000, it just went to crap.

Thanks for doing what you do Olaf and helping all us pro/e guys with justifying good hardware. Really appreciate it.

Steve


Terms & ConditionsCookie settingsAccessibility statement