ATi's Mobility Radeon 9000: DX8 for Notebooks

DirectX 8 For Notebooks

Only a short while ago, ATi unveiled Radeon 9700 , currently the most advanced 3D chip for desktop systems. This chip brings highly impressive features to your screen, already meeting the future standards of Microsoft’s upcoming DirectX 9.

Things are all different in the notebook arena. Here, 3D graphics aren’t as important and, at the same time, are much more difficult to implement. It still seems somewhat disappointing that since the inception of the first full-blood notebook 3D accelerator chip (NVIDIA’s GeForce2 Go ) almost two years ago, the capabilities of notebook 3D graphics have remained at DirectX 7 level. The current 3D-solutions ATi Radeon 7500 as well as NVIDIA GeForce4 Go may have been launched quite a while after the release of DirectX 8, but none of the two come with programmable vertex or pixel shaders. Today, notebooks are finally able to catch up with their desktop nephews. The launch of ATi’s Mobility Radeon 9000 opens the world of DirectX 8 features to the notebook user.

The Specifications Of Mobility Radeon 9000

Mobility Radeon 9000 (code name ’M9’) is not a dedicated notebook design, but derived from the desktop product Radeon 9000, ATi’s latest value offering. Therefore, the technical specifications of Radeon 9000 aren’t any different for the desktop product either. The desktop version may not require the power saving features, but that does not mean Radeon 9000 and Mobility Radeon 9000 would be different chips. They are completely identical, while of course the notebook versions are specially picked chips that ensure reliable operation at low voltage and power consumption. In the same way that Radeon 9000 introduced new standards to the value desktop market, bringing programmable vertex and pixel shaders to the lower price range of graphics cards, Mobility Radeon 9000 finally makes notebooks capable of running DirectX 8 games with all their impressive 3D-effects.

Before I will simply list all of the relevant features of Mobility Radeon 9000, I suggest you have a look at the article about Radeon 9000 Pro , to learn all about the chip behind ATi’s new notebook graphics solution.

List of Features :

36 million Transistors, 0.15µ-process Core Clock 240 - 250 MHz, Memory Clock 400-440 MHz (both clocks depend on implementation) Theoretical Fill Rate 960 - 1000 Mpixel/s Memory Bandwidth 6400 - 7040 MB/s AGP 8x support Two DirectX 8.1 compliant programmable vertex shaders DirectX 8.1 compliant programmable pixel shaders Four pixel rendering pipelines One texturing unit per pixel pipeline Hyper-Z II Memory Bandwidth Optimization Smoothvision = Super Sampling FSAA Powerplay - ATi’s power management (not implemented in first shipping notebooks !) Integrated MPEG2 (DVD) decoding units, like iDCT, motion compensation, hardware sub-picture decoder, adaptive de-interlacing 400 MHz RAMDAC, 165 MHz TMDS transmitter, integrated video-out Multi-display support with ATi Hydravision, allowing up to three displays at the same time (notebook panel, CRT, television)

The Test System

We received our Mobility Radeon 9000 sample in the form of a notebook. Unfortunately, we are not supposed to disclose the brand of the notebook, but we can tell you that it was equipped with a 2 GHz Pentium 4, i845M chipset, 512 MB of DDR-SDRAM and a 1600x1200 TFT screen. We used driver revision 7.75 for the tests, which do NOT include ATi’s ’Powerplay’ power management. ATi told us that the first shipping notebooks with Mobility Radeon 9000 will not have Powerplay enabled, because it will take some more time until the Powerplay software for the new chip has been fully developed and then validated by the OEMs.

The Mobility Radeon 9000 within the notebook was clocked at 240 MHz, and the memory clock was 420 MHz. Thus the theoretical fill rate was 960 Mpixel/s and the memory bandwidth was 6720 MB/s. The core and memory clock speeds found in the notebooks that will use Mobility Radeon 9000 might vary from those settings, because the OEM will set them according to power requirements and thermal envelope of the notebook design.

The Contestants

We were lucky in that we could run all the other chips on virtually the same kind of notebook platform as well. Mobility Radeon 9000 was up against three opponents.

ATi Mobility Radeon 9000

ATi is using only 4.5 ns memory, which is why the memory clock is a mere 420 MHz.

ATi Mobility Radeon 7500, the predecessor

NVIDIA GeForce4 Go440, the current competitor

NVIDIA’s GeForce4 Go440 uses 3.6 ns memory, which is clocked at 440 MHz.

NVIDIA GeForce4 Go460, the new competitor with increased core and memory clocks

The upcoming GeForce4 Go460 comes with 2ns memory, allowing a whopping memory clock of 580 MHz.

We ran Mobility Radeon 7500 at a core clock of 280 MHz and a memory clock of 400 MHz. GeForce4 Go440 was operated at a core clock of 220 MHz and a memory clock of 440 MHz. Finally, the other new kid on the block, internally dubbed ’NV17M Pro,’ is clocked at 275 MHz for its core, and at a whopping 580 MHz for its memory.

Benchmark Setup

Here’s a comparison table of all the systems :

Swipe to scroll horizontally
ProcessorIntel Pentium 4M 2 GHz
Memory512 MB DDR-SDRAM
ChipsetIntel 845M
Display1600x1200 TFT
Battery66 Whr
Graphics Subsystem
Row 6 - Cell 0 ATi Mobility Radeon 9000ATi Mobility Radeon 7500NVIDIA GeForce4 Go440NVIDIA GeForce4 Go460
Core Clock240 MHz280 MHz220 MHz275 MHz
Memory Clock420 MHz (DDR)400 MHz (DDR)440 MHz (DDR)580 MHz (DDR)
Graphics Memory64 MB64 MB64 MB64 MB
Driver Version7.756.13.10.605236.8036.80
Theoretical Fill Rate960 Mpixel/s560 Mpixel/s440 Mpixel/s550 Mpixel/s
Peak Memory Bandwidth6720 MB/s6400 MB/s7040 MB/s9280 MB/s

You can see that while ATi’s chips are ahead in theoretical fill rate, they lag behind NVIDIA’s products when it comes to memory bandwidth. 3D performance depends much more on memory bandwidth than on theoretical fill rate numbers. That is why the two NVIDIA chips can easily keep up with their ATi counterparts when it comes to the actual frame rates.

Direct3D Benchmarks Mad Onion’s 3DMark2001SE has become a de-facto standard these days. We use it for the overall result as well as the fill rate number. AquaNox : We used the retail version of AquaNox and patch 1.17. Comanche 4 Demo Benchmark : We ran the new version Dungeon Siege : The official benchmark patch was used and the file ’system_detail.gas’ was adjusted to add the configuration for Mobility Radeon 9000. Max Payne : We ran the demo ’Final Scene 1’ for the benchmark.
OpenGL BenchmarksQuake 3 Arena is another de-facto standard, even though it’s very old by now. We used good old ’demo001’ and version 1.17. Jedi Knight II : We used the retail version and the ’jk2ffa’ demo. Serious Sam : We ran the retail version and the ’Valley of the Jaguar’ demo.
Anti-Aliasing or Anisotropic Filtering Benchmarks
I decided against benchmark runs with anti-aliasing or anisotropic filtering enabled, because the scores we got with both features disabled were already so low that it would have been unrealistic to think that anyone would actually play games on his notebook with those features enabled.

Benchmarking Screen Resolutions
Most notebooks come with screen resolutions of either 1024x768 or 1600x1200. It’s always preferable to run 3D Games in the original resolution of the display panel, as it avoids artifacts due to interpolation. We therefore tested at 1024x768 and 1600x1200 only. In both cases we used 32-bit color.

3D Benchmark Results


Of all the benchmarks, 3DMark2001SE is the one where Mobility Radeon 9000 is furthest ahead of the competition. The reason why is due to MR9000’s support of DirectX 8.1. It allows it to run the ’Nature’ subtest of 3DMark2001SE, which is something that all the other cards are unable to do. Still, those numbers don’t really reflect gaming performance.

At 1600x1200, NVIDIA’s upcoming GeForce4 Go460 is reaching the same score as ATi’s new mobile 3D chip, even though GeForce4 Go460 is unable to run the DirectX 8 tests of 3DMark2001SE. This hints in the direction that the performance of Mobility Radeon 9000 might be lagging behind NVIDIA’s upcoming chip at this high resolution.

A look at the fill rate scores of the four contestants is somewhat sobering. While Mobility Radeon 9000 may still reach the highest score in single texturing, it’s only third best in the multi-texturing test. You can see that GeForce4 Go460 looks quite strong here, too.

Quake 3 Arena

In good old Quake 3 Arena, Mobility Radeon 9000 encounters another defeat. Due to its single texture unit per rendering pipe, it cannot score the high results of the upcoming NVIDIA chip in this game, which may not be complex but uses a lot of multi-texturing.

Jedi Knight II

In Jedi Knight II, Mobility Radeon 9000 may have a tiny lead at 1024x768x32, but at 1600x1200x32, GeForce4 Go460 beats it. All in all, the difference in scores between the four cards isn’t exactly huge in this game, which is based on the Quake 3 engine. The 1600x1200 results nicely reflect the fill rates of the cards.

Dungeon Siege

Microsoft’s Direct3D RPG displays similar results as Jedi Knight II. The scores are all pretty close, but Mobility Radeon 9000 has a slight edge. Again, we see nothing spectacular.


AquaNox is an actual DirectX 8 game, so all cards except for Mobility Radeon 9000 are running in some kind of compatibility mode. Again, we don’t see any impressive lead of Mobility Radeon 9000.

Comanche 4 Demo

The scores in the new Comanche 4 Demo are another disappointment. While at 1024x768x32 the results seem to be all the same, Mobility Radeon 9000 seems to have some kind of problem at 1600x1200x32, where the NVIDIA cards are clearly in the lead.

Serious Sam

Finally, Serious Sam is a game where Mobility Radeon 9000 can show its muscles. The NVIDIA chips are lagging behind their competitors from ATi in both resolutions.

Max Payne

The coin turns once more in Max Payne. While all chips score pretty similar results, GeForce4 Go460 wins in both resolutions.

Power Consumption Tests

We wanted to be particularly thorough with the power consumption tests, because after all, we are talking of notebooks here. Luckily, we were able to test all chips on virtually identical platforms. Due to the fact that Mobility Radeon 9000 does not yet come with ATi’s ’PowerPlay’ power saving feature enabled, we wanted to be fair and ran all chips at their normal ’full performance’ settings. We split the benchmarks in three.

First, we did a battery rundown while a perpetual slideshow was running, to simulate normal work on the computer, like e.g. giving a PowerPoint presentation to a customer. In this case, the system is idle most of the time.

Many people use their notebook as DVD-player when they are traveling. Our second battery rundown was while playing back a DVD (Castaway) in full screen mode. In this case, the graphics chip, as well as the CPU, has to do quite a bit more work than in case of the slideshow. Hence, the rundown times are shorter.

Finally, we ran a perpetual 3D gaming demo and took the time until the battery was empty. In this case, the CPU, as well as the graphics chip, has a lot of work to do, so the power consumption is significantly higher than in the above two cases, and the battery rundown times are much shorter.

Slideshow Battery Rundown Test

I was very impressed to see that Mobility Radeon 9000 uses just as little power as its predecessor Mobility Radeon 7500. The two NVIDIA chips are only a little bit behind, with the upcoming GeForce4 Go460 running out of power only four minutes sooner than M9.

DVD Playback Battery Rundown Test

Finally Mobility Radeon 9000 is able to shine. With the new ATi notebook chip you can watch DVD movies about 15 minutes longer than with any of its competitors. GeForce4 Go460 is the clear loser in this comparison.

3D Gaming Battery Rundown Test

If you want to play 3D games on your unplugged notebook at full performance, you are also best off with ATi’s new mobile chip, because, while it may only have a tiny lead in terms of battery rundown time, it delivers the best frame rates for that duration.

Bottom Line

ATi is of course proud to be the first 3D chipmaker that can offer a notebook graphics solution that supports DirectX 8. NVIDIA lost this race just as much as it lost the race against Radeon 9700 Pro in the desktop arena. However, what does the owner of a future notebook with ATi’s new Mobility Radeon 9000 really get ? As long as the majority of games are running just fine on DirectX 7 hardware, the advantage of Mobility Radeon 9000 is minimal. ATi should have made sure to give its new mobile chip more memory bandwidth to bring up real-world fill rate, because this is still the number one requirement of 3D games running on notebooks. The benchmark results speak a clear language. Mobility Radeon 9000 is far from an impressive 3D performer.

It is true that NVIDIA’s yet unreleased GeForce4 Go460 was able to spoil the show, as it can deliver frame rates that are either extremely close to, or sometimes even slightly better than those of Mobility Radeon 9000. We will see which of the two chips will be available first.

If you compare the chips on a power consumption level however, Mobility Radeon 9000 is the best offering out there right now. This will most likely ensure that the new mobile ATi chip will earn the majority of design wins in the near future. However, the 3D benchmark scores of Mobility Radeon 9000 are disappointing, there is no denying that.

NVIDIA will try its best to catch up soon. It won’t take long for NV18M’s release, something like GeForce4 Go460 with AGP 8x support. Only a short while later, and probably before the end of this year, there will be NV28M, a mobile derivative of GeForce4 Ti 4200, which will probably beat Mobility Radeon 9000 in 3D performance, but lag behind it in terms of power consumption. What NVIDIA is really waiting for is the NV3x family of chips. In the Spring of next year we will see a notebook chip in 0.13 micron process with DirectX 9 support. Until then, NVIDIA will have a hard time to compete with ATi in the notebook arena.

For now, Mobility Radeon 9000 is the best 3D solution for notebooks, but its lead in terms of 3D performance is, unfortunately, disappointingly small.