SolidWorks 2009 running on a FirePro v3750 vs Radeon 4850

I have a video capture from the SolidWorks World 2009 conference last week that compares a top-of-the line Radeon consumer card (on the left), vs the low-end FirePro 3750 professional (on the right)  on SolidWorks 2009 performance.  As you can see in the video, the FirePro leaves the Radeon in the dust. Obviously, its the drivers support for OpenGL (specifically VBOs) and their SolidWorks 2009-specific optimizations that give the FirePro card the impressive edge.

With both at the same $200 price, and since I am not a gamer, the FirePro seems the obvious choice.

Supermicro X8SAX produces fastest 64-bit professional graphic scores using FirePro v8700

IT Examiner has a second review out on the Supermicro X8SAX Workstation - s single socket CPU (Intel Core i7) mainboard that has the power of some dual Xeon systems. This new review assesses graphics performance under Windows XP Professional 64-bit SP2 using the FirePro v8700. MAXON CINEBENCH 10 and SPECviewperf 10 results show some of the fastest graphics performance the authors have ever seen from a single socket CPU.  From the article “64-bit mode normally sees a slight hit in performance in the graphical field of approximately 5% - 8%.  Today we do not see this, as the carefully constructed driver from ATI nearing full maturity is hitting home hard and fast and the power of the FirePro V8700 is coming to the fore.”

Review of i7 X58 motherboard reveals interesting comparison of FirePro V8700 to Quadro FX4800

3DProfessor has a great review of the new Supermicro X8SAX workstation motherboards that feature the Intel i7 CPU and X58 chipsets running at 3.2 GHz under WinXP.  These motherboards are clearly aimed at the professional markets such as geophysicists, designers, scientists, engineers, Medical Imagery and other technical professionals.

The review itself is interesting, but equally interesting is that performance was tested using two workstation graphics accelerators:  ATI’s FirePro V8700 and NVIDIA Quadro FX4800.

The relevant comparison is the SPECapc for SolidWorks 2007 and SPECviewperf 10 for the FirePro 8700  and the Quadro FX4800 .
What you will basically see is great performance from both cards with almost identical performance in SolidWorks 2007 (I wonder about 2009 with VOBs!) and varying winners in different SPECViewperf tests (where higher scores are better).

Great results from both accelerators is the takeaway in terms of boards.  Great support for EEC memory is the takeaway in terms of the Supermicro motherboard.

FirePro V8700 review on HotHardWare - performance comparisons in OpenGL, DirectX, GPGPU and gaming

HotHardWare has a detailed review of the FirePro V8700 and compares it performance-wise in OpenGL, DirectX, GPGPU and gaming (yes gaming!) to previous top-of-the-line FireGL cards as well as QuadroFX 1700, 3700 and 5600 cards.

Quick Summary: "Excellent performance increases, very low noise, flexible outputs at a price point of nearly half that of ATI's prior generation high-end product." ... "FirePro is a massive shot in the arm for the product line and delivers on almost every front. ATI has not only addressed every major concern we had about their last generation FireGL V8600-series, but they have enhanced their workstation-class product offerings in other ways as well, while keeping the price point in check. The result is a highly refined product which feels properly tuned to the workstation market."... "The FirePro V8700 is not perfect, of course. ATI still hasn't nailed down their multi-GPU support as well as Nvidia has, nor do their drivers have as much polish as Nvidia's in our opinion (it close, though)."

3D Professor writes on Intel’s i7 965EE with ATI’s FirePro V8700

3D Professor has run a series of performance tests on the next-generation Intel Core-i7 architecture with the FirePro V8700.

Here’s the gist:

i7 CPU and X58 Mainboard from Intel -  The overall performance of the mainboard has been superb, and has many options for upgrades at a later date.  But for now its more than adequate for most.  The new triple channel memory controller most certainly has the required bang needed.  The unequivocal evidence shown within proves just how much raw performance is hidden under the hood.  Therefore for those of you wishing a merry Christmas with a new build - then look no further as Intel has done the job.

ATI FirePro V8700 - The performance shown today has been well worth its wait.  The results shown within have been extra-ordinary to say the least.  SPECviewperf 10 scores have never been so high from a single socket CPU.  This card most certainly has overtaken its predecessor in many ways, faster memory, the aid of 800 unified shaders - this will help many in complex drawings.  Life in the professional graphics card market has just shifted up an a incredible gear.  Today’s outing has been the tip of the iceberg.  Cost and availability of the ATI FirePro V8700, currently the card fires in at an MSRP of approximately $1499 this is a very respectable price bracket for something which performs so well in all areas.

Video capture of FirePro V5700 vs Quadro FX 1700 with CATIA tested at Develop 3D Magazine

Greg Corke and Martyn Day of Develop 3D magazine recently supervised a series of side-by-side performance benchmarks in CATIA comparing the FirePro V5700 and the Quadro FX 1700.  The models were all high-polygon, high complexity (i.e. requiring lots of culling in shaded view) from CATBench. Both cards include 512 MB RAM and both are running on identical HP xw6600 workstations. The AMD folks were on hand with a digital camcorder to capture the results.

The test runs through three models (1.5 million, 3.9 million, and 3 million polygons) and has each rotate, tilt, and zoom in shaded, shaded plus outline, and outline-only views.

Interesting notes:  For low-polygon, low-complexity models, the Quadro FX performed as well as or better than the V5700. But as model complexity increased, the v5700 really outpaced it. I am guessing that this is due to the optimized OpenGL VBO support on the FirePro line which enables the graphics card to store the geometry on the card and even update it without having to push the geometry back down the bus to the CPU.

Tom’s Hardware compares 8 FireGL and 8 Quadro FX cards for the pro market

Tom's Hardware today ran through a rather comprehensive comparison of workstation graphics cards. What makes comparison so interesting is the comprehensiveness - 8 current FireGL and 8 Quadro FX cards (although not the new FirePro cards announced at SIGGRAPH).

Cards were compared on 3D Studio Max 9, Solidworks 2007, Maya 6.5, and Viewperf 10 for CATIA, Ensight, Pro/Engineer, UGS Teamcenter Visualization Mockup & UGS NX.

Tom's Hardware Conclusion: "The Nvidia Quadro FX 5600 may wear the performance crown, but the AMD FireGL V7700 is working at chopping off the legs of Nvidia’s throne. The FX 5600 takes first place in many benchmarks. However, it can’t claim to be the undisputed leader - the FireGL V7700 wins in too many categories."

But as I wrote in a previous post on benchmarking, the best benchmarking tool is to try the cards out to see which performs better in your environment, with your workflow and files (you should check the return policy for particular web vendors to make sure you can return a card).

1. Published benchmarks are certainly a good starting place for your own evaluation.
2. Recommendations from colleagues can also be useful. But make sure they have actually tested the alternatives (i.e. they are not just speaking as a fanboy), that you know what version of driver they tested, and that you understand what their workflow is like relative to your own.
3. The best benchmark is for you to compare the cards yourself, using your data and your workflows, if possible.

What is the best benchmarking tool for graphics accelerators?

I recently read a question on CGSociety forums on how to evaluate graphic card using benchmarks. Let me share an edited version of the question and then provide what I consider the most realistic and valuable answer.

Question - How do benchmarks reflect real life experience?
Every review I’ve seen shows rather dramatic advantage for the FireGL over the Quadro in Maya’s benchmark. But everyone I have asked, has only used a Quadro and so that is what they recommend. How do benchmarks reflect real life experience? The prices for the two cards are pretty similar on the Net, so if I am comparing them in terms of performance in Maya, how should I choose?

Answer:
Benchmarks are a single metric for evaluating any product, including graphics cards. They test only what they are designed to test. It is virtually impossible to create a benchmark that will cover all possible user cases. Benchmarks are based on specific workloads and actions to run. And while you can look specifically at those benchmarks that best align with what you do on a daily basis, the workloads and tasks are never truly personalized.

The best benchmarking tool is to try the cards out to see which performs better for your particular needs (you should check the return policy for particular web vendors to make sure you can return a card). There is a lot of brand loyalty out there for many products, including graphics cards, which typically is not based on any kind of comparison against the competition. What may have been true for accelerators from a year ago, is not likely true today, as the drivers are constantly enhanced and optimized.

With respect to published benchmark scores, FireGL cards have shown a performance advantage over Quadro in SPECViewperf for a while now, particularly with Maya. It may be that the graphics architecture for FireGL is better suited to run Maya. Or maybe the FireGL graphic drivers are simply better tuned. Even though others may be recommending to go with Quadro, it would be unwise to completely ignore documented benchmarks performance advantages. The difference just means you need to look more closely.

1. Published benchmarks are certainly a good starting place for your own evaluation.
2. Recommendations from colleagues can also be useful. But make sure they have actually tested the alternatives (i.e. they are not just speaking as a fanboy), that you know what version of driver they tested, and that you understand what their workflow is like relative to your own.
3. The best benchmark is for you to compare the cards yourself, using your data and your workflows, if possible.

An informed decision is always the best decision.

Tech specs alone don’t tell the full story - consumer vs workstation accelerators

If you just look at the tech specs of the top of the line $299 Radeon HD 4870 consumer card and try to compare them to something like the mid-range $1,099 FireGL V770 workstation accelerator you might be a little perplexed.  On paper it hard to see the differences clearly.  Both offer 512MB of memory, PCI Express, Direct X 10.1 with full Shader Model 4.0 support, DisplayPort, dual-link DVI supporting up to 30-bit digital displays at 2560x1600 . The Radeon also has 800 stream processors vs the FireGL with 320 processors. And the Radeon has a bunch of support for audio control to boot!

So is the difference between the midrange workstation graphics accelerator and the high end consumer card, just a lot of marketing hype. Are you just paying a premium to have the words “professional” and “workstation” in the marketing literature?

It turns out its a little more complicated then just the raw specs. Gamer cards push game graphics around fast. This often means high memory bandwidth for texturing, fast full screen anti-aliasing, and fast shader performance. Workstation cards typically are better at line-antialiasing, clip planes, an order of magnitude better with high polygon count work, much better working with mutiple windows or overlays, and are highly optimized and tuned for OpenGL performance. (The vast majority of modeling, CAD, and high-end visualization applications opt for OpenGL rather than DirectX, the latter being the common API for gaming-level graphics on the desktop [but not mobile].)

High end gaming cards focus on pure speed (where close approximation is sufficient)  while high-end workstation cards specialize in extreme accuracy and precision. They are incredibly accurate with FSAA and sub-pixel precision. Workstation graphic cards also have other features such as overlay plane support, multiple clipping planes, many layers of transparency, application performance optimization auto-detect, 3D stereoscopy, and full support for OpenGL (currently 2.1, but 3.0 when it is released).

Workstation cards also include technical support (i.e. you can actually get a trained workstation SE to take your call) and importantly, the accelerators stay on the market and are supported (including parts and replacement) for at least 3 years.

Finally the FireGL drivers are ISV certified, meaning that independent software vendors like Dassault, Bentley, SolidWorks and Autodesk certify on a quarterly basis, that the drivers perform well with specific versions of their programs. Certification is a massive multi-stage process and ensures accuracy for mission critical projects. It also yields substantial performance benefits (e.g. FireGL driver technical collaboration and certification with Dassault in 2007 resulted in 10-fold performance improvements in CATIA v5).

Oops - one more:  FireGL drivers support Linux, as well as Windows XP/Vista in 32 & 64 bit flavors. (A 33% SPECviewperf 10 boost under Linux was recently announced.)

While the gaming market does get the vast majority of the press and discussion online, the workstation market is equally as important to engineering, visualization and design companies. The general take away is that a solid general purpose / gaming card like the Radeon HD 4870, will usually be good enough for most hobbyist 3D animators/illustrators and CAD tinkerers.  But when you get into professional CAD work with millions of polygons and rigorous standards for precision, high end video fx, medical visualization where detail means life or death, or you work in a corporation where reliability and support are required, workstation accelerators more than pay for themselves.

Advances in real time medical visualization using the GPU

I was doing a little research for a project on medical visualization when I came across a video and presentation about using the GPU for real-time medical visualization. The PDF presentation covers combining isosurface extraction and direct volume rendering to create a demo for the Visible Human project. 

But it is the video that is totally impressive. It demonstrates using MRI and CT Scan data, letting the user manipulate the views and slices, then creates complex iso-surfaces to visualize various tissues such as bone, muscle and skin. (note: if you have a current FireGL accelerator and Vista you can actually run the application demo!).