Review Index:

NVIDIA Quadro 5000 2.5GB Fermi-based Professional Graphics Review

Author: Ryan Shrout
Manufacturer: NVIDIA

Testing Methodology, System Setup and CineBench 11/10

NVIDIA Quadro 5000

Looking at the GPU-Z information on the Quadro 5000 we see that the speeds on all of the clock speeds are relatively low compared to the consumer offerings.  While the core speed of 513 MHz, shader speed of 1026 MHz and memory speed of 2.6 GHz might seem fast, the GeForce GTX 465 had reference clocks of 607 MHz, 1215 MHz and 3.2 GHz respectively.  With an average clock drop of 18% on the GPU and 23% on the memory, that does help explain the power efficiency of the new Quadro 5000 and should help indicate relative performance levels to consumer cards for gaming.

Testing professional graphics board is a much different beast that our
reviews on consumer graphics solutions that focus nearly completely on
gaming.  The user interested in a Quadro or FirePro card MIGHT be
curious how the card performs in gaming, but in reality the decision to
buy a card of this cost is based on other application performance: 3D
modeling, CAD and design tools. 

By far the most interesting tests for us with this kind of product
are the SPEC series of benchmarks.  SPECviewperf has long been a staple
for evaluating professional level application performance and the newest
version enables us to test a multitude of results.  Applications like
Maya, 3D Studio Max, SolidWorks and UGS NX are simulated as are several
others; the benchmark allows us to easily test how the graphics cards
help in scaling with multisampling as well as multi-threaded

For this review we are not only including results from SPECviewperf 10 but also the newer version 11 of the testing suite that removes some older test scenarios, adds some new ones, increases the standard resolution and thread counts and just generally does a better job of simulating the latest professional software on the market. 

Also, we have
included the OpenGL portion of the CineBench 10 and 11 rendering
benchmark and
even given 3DMark Vantage a whirl for those of you really interested in
how 2GB of frame buffer affects performance.



System Setup


Intel Core i7-965


ASUS P6T6 WS Revolution


Corsair 3 x 2GB DDR3-1333



Western Digital VelociRaptor 300GB






NVIDIA Quadro 5000 2.5GB

AMD FirePro V8800 2GB

AMD FirePro V7800 2GB

NVIDIA Quadro FX 4800 1.5GB



 NVIDIA Quadro 258.98

AMD FirePro 8.723

Power Supply Corsair Professional AX1200 PSU



DX11/ DX10

/ DX9c / OpenGL



Windows 7 Ultimate 64-bit

There are some important price considerations to look at with these four cards.  First, the AMD FirePro V8800 currently is selling for about $1300 while the V7800 sells for under $800.  The current cost on a Quadro FX 4800 is $1549 (though I would expect that to go down now) and the estimated price for the new Quadro 5000 is $2249 - noticeably higher than the best option available from AMD.  For the professional graphics world, a difference of about $1000 is probably pretty easy to justify if there are performance advantages to worth getting, but they had better be substantial.

UPDATE: We heard from AMD literally hours before this review was to be published about new drivers for the AMD FirePro cards that would:

  • More than 50% increase in performance for ATI FirePro V8800 on the 3ds Max subset of the SPECviewperf 10 benchmark, as well as a 40% gain on the Catia subset, a 20% gain on the Pro E subset and more than 20% gain on the overall composite.
  • Across the board performance increases for the entire next-generation family of ATI FirePro graphics on the Lightwave subset of the new SPECviewperf 11 benchmark, with the entry-level ATI FirePro V3800 achieving a 62% performance gain and the ultra-high ATI FirePro V8800 achieving an 81% gain, compared to the previous driver.

  • Across the board performance increases for the entire next-generation family of ATI FirePro graphics on the SolidWorks subset of SPECviewperf 11, with the entry-level ATI FirePro V3800 achieving a 43% performance gain and the ultra-high ATI FirePro V8800 achieving a 63% gain compared to the previous driver.

While I haven't had time to validate these claims, I wanted to note the information from AMD before progressing to our results using the previous driver. 

CineBench 10 and 11

rendering benchmark based off of the Cinema 4D engine is a terrific
indicator for multi-threaded processing.


While the CineBench 10 result is basically pointless at this time the CineBench 11 results show the AMD cards soundly thwamping both the new and old Quadro offerings.  NVIDIA did alert me to this before testing started and said they had a driver fix for the Fermi-based Quadro cards on the Cinema4D engine coming later in the month.