*NEW* SPECviewperf® 9 and Conclusions

This is a very new benchmark from SPEC the download is very big indeed (622MB) and we would suggest using a high speed DSL connection.   We therefore move onwards and a brief summary of paragraphs below which are taken directly from SPECviewperf ® 9 information release and it gives added insight to the updated version.

A significant change in SPECviewperf 9 is the use of glDrawElements instead of glArrayElement within the restructured Maya viewset. With the change, SPECviewperf 9 follows the same OpenGL command stream as Maya 6.5, handling graphics data in the same way as the actual application.

SPECviewperf 9 includes the following new features that enable it to more closely mirror typical application performance:

• larger, more complex viewsets that place greater stress on graphics hardware;

• memory and list allocation improvements that allow data to be reused and shared in the same manner as within actual applications;

• better compression, enabling the inclusion of larger viewsets; and

• mixing of primitive types and graphics modes, helping to ensure that optimizations for a viewset will be reflected in real-world performance.

“This new version represents tremendous strides in accommodating complex data sets and making SPECviewperf results more closely represent real-world performance,” says Ian Williams, SPECopc chair. “This is still a synthetic benchmark, but one with a distinctive ability to accurately capture and codify the performance of graphics operations within popular applications.”

New viewsets in SPECviewperf 9 are based on traces of UGS Teamcenter Visualization Mockup, a conceptual design application, and UGS NX 3, a leading CAD/CAM application. Both new viewsets use very large and complex models – up to 11 million vertices in the Teamcenter Visualization Mockup viewset, and 30 million vertices in the NX 3 viewset.

SPECviewperf measures the 3D rendering performance of systems running under OpenGL. The SPECopc^SM project group has worked with independent software vendors (ISVs) to obtain tests, data sets and weights that constitute what is called a viewset. Each viewset represents the graphics rendering portion of an actual application. The ISVs that develop SPECopc viewsets have provided percentage weights for each test for which a performance number is reported. ISVs have defined these percentages to indicate the relative importance of a test within the overall application.

Finally, the SPECopc^SM project group's SPECviewperf 9 is totally new performance evaluation software, with major code rewrites, two new viewsets (tcvis-01 and ugnx-01), and a viewset with completely restructured code (maya-02). Since the SPECviewperf source and binaries have been upgraded to support changes, no comparisons should be made between past results and current results for viewsets running under SPECviewperf 9.

Conclusions

We see once more another remarkable card appearing out from the ATI stable, this card packs a punch in all areas.   The only downside we could see was size of this card, this should certainly not deter the user as it has potential to wreck havoc within its price positioning and performance in the market place.  

With this aside we have seen the recent announcement of the new mainboard platforms from both Intel and AMD – how will each card perform upon each platform.  Time will tell and what we can say to you is that we have some interesting reviews already being planned to show off cards from all camps on both AMD and Intel platforms.   This will give a whole different look to each graphic card and mainboard with the current results we have already seen.

Within SPECviewperf ® 8.1 which many of you are used to, we see an upwards trend from the results in the card which is to be expected.  In SPECviewperf ® 9 and the tightening up of the viewsets and implementation of glDrawElements we see much different results with the exception of maya-02 that stomps ahead.    As time allows we will be bringing you results from the Quadro range so that you may have direct comparisons to correctly gauge the cards from each other in Performance v Price.  

In each application as would be expected there are marked improvements over the FireGL V5200 which we displayed to you a few weeks back.   This should be so with the FireGL V7200 new 512-bit Ring bus architecture and near double the Memory Bandwidth over the 5200’s 256-bit Ring Bus.    Noticeably in the new Cinebench 9.5 benchmark the results sneaked up nicely as too with SPECviewperf ® 8.1 we see considerable advancements across the range of benchmark viewsets.   SPECviewperf ® 9 we will be producing very soon results from all the range of cards recently shown so that the reader can better understand the results.

Current Estimated RRSP $1067 USD, £ 599UKP, €838 EURO (before Taxes and Shipping), cards are shipping already and are in high demand.   Distributors are coping with demand from SI and VAR’s.                     

ATI certainly have proven with their new whole range of professional workstation cards that the race is on now.   From what we have seen over these past few months, this has been a resounding upgrade from ATI’s previous generation of professional workstation cards.    

But what’s next we ask ourselves has anyone from ATI or NVIDIA got something left up their sleeves?     Computex is just around the corner so let’s just see what it brings!

Page 1 - Introduction
Page 2 - System Configuration and Benchmarks Used
Page 3 - Benchmark Results
Page 4 - SPECviewperf® 9 and Conclusions