A New Contender: NVIDIA GeForce4 Go
Last month we took at look at ATI’s new Mobility Radeon 7500 and we were excited about the performance that it had to offer (The Mobility Radeon 7500 – NVIDIA and ATI Go Head to Head In The Mobile Market). At almost the same time we released that article, NVIDIA launched it’s new GeForce4 product line at a posh affair in San Francisco. The new GeForce4 Go is, of course, a member of the new GeForce4 product line.
The GeForce4 product line was perhaps one of the worst kept secrets that we have seen in a long time. Most consumers knew many of the details about the product before the launch, so, while it perhaps didn’t make the biggest splash in history, it was a very exciting move forward for NVIDIA nonetheless. We covered the launch of the GeForce4 desktop product line in the following article: PC Graphics Beyond XBOX – NVIDIA Introduces GeForce4
Although we have all heard most of the details and seen the performance of the GeForce4 desktop products, still, as of press time, little has been written about the GeForce4 Go, which was also known as the NV17M. Many of the issues regarding the absence of information about the GeForce4 Go have been due to the lack of review units available for testing. THG has been affected by this problem, as well. It was never a question of if we would review the GeForce4 Go, but a question of when a unit would be made available to us so that we could review the GeForce4 Go.
Another issue that has surrounded its launch has been the lack of design wins for the GeForce4 Go. By our count, NVIDIA did not really gain that much ground (so far) in new OEM design wins. Currently, only Dell and Toshiba are selling units with the GeForce4 Go in them. NVIDIA did pick up some additional design wins by adding the new Acer TravelMate 630 and WinBook n4, however, they both use GeForce2Go, rather than GeForce4 Go, designs. However, as of this article, you can only find the GeForce4 Go in the Dell Inspiron 8200, Dell Latitude C840, Toshiba Satellite 5105-S507, Toshiba Satellite 5105-S607, and Toshiba Satellite Pro 6100. Industry sources seem to indicate that you can expect to see additional OEM wins for the GeForce4 Go in other Pentium 4m units, once OEMs announce Pentium 4m notebook product refreshes later this year. It will be interesting to see how many additional units we be announced with the GeForce4 Go at this time. At least for the time being, it will be easier to find units with the ATI Mobility Radeon 7500.
When NVIDIA launched the GeForce4 Go, the only unit available with the GeForce4 Go in it was the Toshiba Satellite 5105-S507, which was the Pentium III 1.1Ghz unit with the 100MHz FSB. Of course, as we explained in depth in our look at the ATI Mobility Radeon 7500, comparing different notebooks on different platforms is a difficult task as the machines and configurations are not the same. In the case of the Toshiba 5105-S507 and the Dell Inspiron 8100, the difference in FSB speed alone makes any comparison pretty much useless, in our opinion.
After much waiting and wondering, we were provided with a final engineering sample notebook from Dell for use in the testing of the GeForce4 Go. As excited as kids at Christmas time, we dug deep into the GeForce4 Go to find out if it is the evolution that NVIDIA would lead us to believe it is.
Is GeForce4 Go More Than A Mobile Of Version Of The GeForce4 MX?
What is the GeForce4 Go? Answering this question is more challenging than it might appear on the surface. Yes, the GeForce4 Go is built on a highly optimized version of the NV17 core that is found in the GeForce4 MX product line, but it offers a lot more than that. The GeForce4 Go is built on technology that was first introduced with the GeForce4 and its NV17 core. The NV17M core that is found in GeForce4 Go is built on a 0.15 micron process, which is quite a bit more compact than the 0.18 micron process that was used in the GeForce2Go. Of course, by shrinking the GPU with this new process, it was possible to ramp up the clock speed while reducing heat.
A close up look at the GeForce4 Go GPU.
One of the biggest byproducts of moving to the 0.15 micron process is that NVIDIA was able to reduce power consumption. When you combine this with the new PowerMizer dynamic power modulation hardware and software technology, you should be able maximize battery life. One of the major complaints about the GeForce2Go was that it had a tendency to be a little power hungry, by current standards.
The GeForce4 Go 256-bit graphics engine is available in five different flavors. Each of these flavors provides different configurations that OEMs are able to select based on need. These configurations help address the OEMs’ need to have a scalable solution for different notebook market segments. Of course, the main differences between these configurations are not only in the packaging, but in the relative performance for each specific market segment. In addition, NVIDIA will continue to market the GeForce2Go to interested OEMs, so don’t expect it to go away anytime soon. Here is a breakdown of the different module configurations for the GeForce4 Go.
Product | GeForce4 440 Go | GeForce4 440 Go 64MB | GeForce4 420 Go | GeForce4 420 Go 32MB | GeForce4 410 Go 16MB |
On-Chip Memory | External | 64MB | External | 32MB | 16MB |
Max Memory | 64MB | 64MB | 64MB | 32MB | 16MB |
Memory Bandwidth | 8.0GB/Sec | 6.4GB/Sec | 3.2GB/Sec | 3.2GB/Sec | 1.6GB/Sec |
Package Type | Discrete | MAP | Discrete | MAP | MAP |
Package Size | 31MMx31MM | 31MMx31MM | 23MMx23MM | 31MMx31MM | 31MMx31MM |
As you can see from the table above, with a package size of 31MM, the chip is still about the size of a normal desktop GPU. The advantage of this is that it has allowed NVIDIA to integrate the GPU core, frame buffer, and memory into the same package, which they call “MAP” (Mobile AGP Package) technology. By using this MAP technology, they are able to fit the GPU and memory into the space that would have previously been occupied by just the GPU alone.
An up-close look at the GeForce4 Go GPU die.
Depending on the configuration implementation by the OEM, the GPU can be run with a maximum clock speed of 220MHz and supports 32-bit DDR SDRAM with a maximum clock speed of 440MHz. The DDR SDRAM supports NVIDIA’s LMA II (Lightspeed Memory Architecture II), which has become one of the big features of the GeForce4 product line. Like most other mobile video solutions, NVIDIA offers AGP 4X with fast write support across the GeForce4 Go product family.
LMA II allows the GeForce4 Go to increase the memory bandwidth between the GPU and memory. Using NVIDIA’s vertex shader acceleration and VPE (Video Processing Engine) results in a mobile graphics model that is even closer to desktop performance. With the higher resolutions found in today’s notebooks, every attempt to optimize the memory bandwidth should, of course, lead to higher performance gains. Although the version of the LMA II that is found in the GeForce4 Go is a stripped-down version of that found in the higher-end GerForce4 desktop product line, it still does offer some advantages and performance gains over the older LMA technology.
How Fast Do You 4Go?
Going fast in the mobile video product space is the result of a combination of many things. That the GeForce4 Go is using much of the technology derived from it’s big brother isn’t a real surprise.
The GeForce4 Go features an integrated transform and lighting engine for texture and lighting (T & L) support, using the GeForce4 Go’s 256-bit graphics engine with 4 texture-mapped, filtered, and lit textels per clock cycle.
You might have already noticed that we have not mentioned NVIDIA’s nfiniteFX technology, which, in NVIDIA branding-speak, means that the card supports vertex shading and offers full DirectX 8.x support. The GeForce4 Go does not include the nfiniteFX-engine, so you therefore don’t get the current pinnacle of graphics technology. This is not really as much of an issue as it might appear on the surface, because currently no one has vertex shading and full DirectX 8.x support for a mobile platform (yet). New software that supports vertex shading has been slow to arrive so far, but it is important to note this for your future reference. So, if you are putting off the purchase of a new notebook because you’re waiting for full hardware support for vertex shading and DirectX 8.x, it is going to be some time before you see it.
Here is an up-close look at the Dell GeForce4 Go 64MB DDR modular board will be found in the Dell Latitude C840 and Dell Inspiron 8200.
In the GeForce4 Go, you will find support for 32-bit color with 32-bit Z/ Stencil buffer support. The NVIDIA Shading Rasterizer (NSR) is able to pump out up to eleven textures and lighting operations per clock cycle. Single pass multi-texturing and cubic environment mapping are also supported. It is ironic that these features have now become so standard in graphics card technology that we often just quickly gloss them over, because in notebook video technology, this is still a rather new addition.
The GeForce4 Go offers full AccuView Antialiasing that has been specially enhanced for the mobile market. The GeForce4 Go version of AccuView attempts to address these issues by being more focused on visual quality, as it is much more noticeable on the higher resolution LCDs found in today’s notebooks. The AccuView Antialiasing technology supports 2X, 4X, Quincunx, and the new 4XS sampling modes. The new 4XS sampling mode is targeted to provide users with visual quality that is more like what you find with desktop PCs. With AccuView turned on you can expect to take a performance hit, though it was not as big as we might have expected. I think previously we questioned the need for more powerful AA technology in notebooks, but, based on the increased need for higher visual quality, it will only be truly useful if we don’t experience too much in the way of performance degradation.
When You Need To Be “Cheap” With The Battery Power, Call In The PowerMizer
We were first introduced to NVIDIA’s PowerMizer technology with the GeForce2Go product line. Many were less than impressed with the power consumption of the GeForce2Go. Of course, we are not sure that proper testing was done to verify the impact of PowerMizer on overall system performance. Power consumption remains a major issue if you wish to use your notebook to achieve true mobility. It is difficult for NVIDIA alone to shoulder the load of this burden when the current notebook designs are ahead of the projected power consumption curve.
When talking about using a notebook on battery power, many talk about the ultimate goal of being able to run full power for four hours. Currently, battery chemistry is still holding us back. Battery technology has evolved, but the kind of big breakthrough that is necessary in order to move battery technology to the next level is still on the horizon. In the meantime, notebooks have found other ways to maximize power usage, while still having to deal with the advances in new notebook technology.
PowerMizer addresses these concerns by allowing you to manually throttle power usage of the GeForce4 Go on the fly, via the PowerMizer control panel. The PowerMizer control panel provides for three settings, which will increase or decrease the power requirements of the GeForce4 Go. These three settings are: Maximum Battery Life; Maximum Performance; and Balanced, which balances both performance and battery life.
A look at the PowerMizer control panel.
It should be noted that, as with the ATI Mobility Radeon 7500, the GeForce4 Go still requires the OEM to offer all of the power management options that the product has to offer. Neither NVIDIA nor ATI can control what level of function the OEM chooses to offer its end users. We have been told that many OEMs choose not to offer these advanced power management options because they fear the possible increase in support costs, due to possible user problems.
The native drivers that we were provided with had the PowerMizer option hidden and not enabled. NVIDIA provided us with a registery key that would enable PowerMizer within drivers for testing. Many OEMs still are on the conservitive side when it comes to the certification of new technology for notebooks. It should come as no surprise that until OEMs have logged a significant amount of testing miles on these new power management technologies, they are going to be slow to add them. We understand that power management can be a cause of stability issues with today’s notebooks, but still feel that giving the user the option to enable or disable them is better than ignoring the existance of the technology. We continue to call on OEM’s to provide full support for all forms of power management technology. As we understand it, Dell, for example, has opted not to support or offer PowerMizer at this time.
When You Need To Be “Cheap” With The Battery Power, Call In The PowerMizer, Continued
In the past, is has been hard to verify these claims of power management. Until now! We were able to use an inline test device to see the actual performance difference between the various PowerMizer settings.
This is the inline power adapter that allowed us to tap in and monitor what benefit or lack of benefit exists across PowerMizer’s various settings in real time.
To test PowerMizer, we placed an amp meter between the AC adapter and the notebook, using our in line test device to monitor the current draw of the notebook while it is operating. The amp meter was located between the AC power adapter and the notebook, which allowed us to montor the total system current draw in real time. During these tests, we removed the battery, so the charging circuitry had no effect on the results. The amp meter allowed us to monitor the amount of current that is drawn by the system amps. All current that is being drawn by the laptop is able to be monitored by the amp meter, as there is no battery installed. With the test adapter installed and battery removed, the system throttles itself back to 1.2GHz just as it would if it were running from battery. To monitor our results, we run the first scene of 3D Mark 2002, which in this case provides an ample opportunity to monitor the system. In order to better understand the effects of PowerMizer, examine the following:
With PowerMizer in the MAX performance setting, the GeForce4 Go operates at maximum current draw. The total system current was 2.1 amps, the AC adapter was 20 volt DC power source. To obtain the total power consumption, you would multiply amps by volts, which gives you the total power consumption in watts. We can then take our total of 42 watts and divide it by 37.4, our total watts in the middle (balanced) settings. We then arrive at a total of 12.3% difference between the maximum and balanced settings.
Max Performance – 2.1A * 20V = 42W – (42 / 37.4) – 1 = 12.3%
With PowerMizer in the middle (balanced) setting, the GeForce4 Go is operating in a balanced mode of current draw. The total system current was 1.87 amps, the AC adapter is 20 volt DC power source. To obtain the total power consumption, multiply amps by volts, which gives you your total power consumption in watts. We then take our total of 37.4 watts and divide it by 33.6, which is our total watts in the max battery setting. We then arrive at a total of 11.3% difference between the balanced and maximum battery life settings.
Balanced (Middle Setting) – 1.87A * 20V = 37.4W – (37.4 / 33.6) – 1 = 11.3%
With PowerMizer in the max battery saving mode setting, the GeForce4 Go is operating in a battery saving mode of current draw. The total system current was 1.68 amps, the AC adapter is 20 volt DC power source. To obtain the total power consumption, multiply amps by volts, which gives you your total power consumption in watts. To understand the difference in the PowerMizer settings, we can now compare max battery to max performance. We take our total of 42.0 watts and divide it by 33.6, which is our total watts in the max battery life setting. This allows us to arrive at a total of 25% difference between the maximum battery and maximum performance setting.
Max Battery – 1.68A * 20V = 33.6W – (42 / 33.6) – 1 = 25%
We have provided a graph to help you compare the three PowerMizer settings:
During the testing process, we were able to verify a reduction in current draw by monitoring readings during various states of use. In the balanced mode, we saw about an 11.3% reduction in required power. In the maximum battery saving mode, we saw about a 25% reduction of power. When operating at a low 1.575 volts, using PowerMizer becomes helpful in gaining additional battery life. Although 25% might sound like a big number, it doesn’t translate into 25% more “real world” battery life.
As we show in our benchmarks, in certain situations PowerMizer has little or no effect with certain applications. PowerMizer does save power, but depending on the applications you use, your mileage may vary.
NVDVD: Another Option That Might Be Coming To A Notebook Near You!
Our test notebook was loaded with the latest version of NVIDIA’s new NVDVD playback software. Although we don’t yet know what systems are going to include this software, we jumped at the chance to have sneak peek at NVDVD.
NVDVD has been in development at NVIDIA for some time, and we have been eagerly anticipating it’s release. A lot of what we found in NVDVD we have seen in other DVD playback software, but it was interesting to see what kind of spin NVIDIA would be able to put on it.
Our test system featured the new nVDVD playback software. We thought the performance was good and it ran well during our testing. It will be up to OEMs whether to include nVDVD with your NVIDIA product purchase.
NVIDIA claims that this is the first DVD playback software that is completely built and engineered from the ground up, specifically for NVIDIA products. This should give NVIDIA an advantage, because NVDVD is designed to take advantage of the features that are unique to NVIDIA platform products.
NVDVD was designed to provide seamless video playback, high-quality audio, and reduced CPU utilization and power consumption. NVDVD supports MPEG-2 video and audio decoding, including support for hardware motion compensation, inverse discrete cosine transformation (IDCT), and inverse quantization (IQ).
NVDVD targets the goal of achieving higher quality output by supporting adaptive de-interlacing and scaling algorithms that are already present on NVIDIA GPUs in order to deliver playback at HDTV-quality resolutions.
NVDVD’s other features include: advanced frame capture; auto-resume; digital vibrance control (DVC); and nView support. We had a chance to test the advanced frame capture and auto-resume feature. Both of these features worked very well, and we found them to be both easy to use and useful.
Depending on the version, NVDVD can include full Dolby Digital support and is able to output this via a digital output if available on your system for external decoding. Of course, this will depend on the compatibility of your sound card.
NVDVD is optimized for performance and is fully Windows Hardware Quality Labs- (WHQL) certified for use with Windows XP, Windows 2000, and Windows ME. NVDVD is optimized to take advantage of MMX, Intel SSE, and AMD 3Dnow.
NVDVD, which will be included free with some systems, and is a great player with a lot of very flexible features. Although it might not be as robust as some of the other solutions available, we found it to perform well and think it is definetely worth considering.
NVDVD can be licensed by OEMs for distribution, but it is up to the OEM. The OEMs will choose to license NVDVD, or not. This of course means that you will find it bundled with some notebooks and not with others. The same will hold true for desktop NVIDIA cards as well, you will find it included with some cards and not with others. We could not confirm if NVDVD will be offered with Dell products at this time. Our questions on this subject were answered with a very solid “No Comment”.
What The Future Holds
Our sample Dell notebook with the 64MB NVIDIA GeForce4 Go.
A question that you may have by now is, “What is a Final Engineering Sample Notebook?”; The easy way to answer this question is that this notebook is the final release version of the notebook before the company starts mass production based on a unit. In the case of this unit, it was neither a Latitude nor an Inspiron. Although you will notice the name Latitude on the outside of the case, it is really a combination of the two notebooks that Dell is releasing, which are the Latitude C840 and the Inspiron 8200. Elements from each of these products made up our test unit. We can learn a lot by looking at this sample, though what we saw in our look at this unit may vary from the final units that Dell chooses to ship.
When first looking at the the Dell Engineering Final Engineering Sample Notebook, right off the bat you will notice that it does, in fact, look at lot like the Latitude C810 and the Inspiron 8100. These observations would not be far off the mark, as the majority of the plastics that make up the outside of this notebook were taken directly from those systems, with very little change. Of course, this is not necessarily a bad thing, but we might like to see Dell make some additional updates to the outside looks in the next revisions of the product.
Our test notebook features the Intel Pentium 4m at 1.7GHz. The unit also features the Intel i845MP chipset. It accepts two SO-DDR SDRAM in a PC2100/266MHz configuration. It can be configured to accept up to 1GB of RAM. Our test unit features use of Dell’s innovative heat pipe cooling solutions for both the processor and the GeForce4 Go GPU. Like the C810/8100, two fans in the back are used to remove heat from the unit. At times, these fans can be a little loud, but they are varible speed and they were not as loud as we have seen on others.
Weighty Power
When discussing the power requirements of our test unit, it is important to note that the normal Latitude/Inspiron 70 Watt (20V 3.5A) power adapter will work with the test unit, but there is a catch. The catch is that the test unit, like the new Inspiron 8200 and Latitude C840, requires the use of the new Dell Latitude/Inspiron 90 Watt (20V 4.5A) power adapter to run at full power. The older power adapters that you may have from your old Latitude/Inspiron series notebooks do not provide enough power for our test unit to run at full speed. This is of course worth noting if you have a heavy investment in mobile Dell products and you are going to add a unit like this to your current IT standards list. The new AC adapter doesn’t add any weight, which is a good thing, but the 1A difference at 20V is really a big deal, as you can tell by the 20 Watt increase in power. The older Dell Latitude C8xx series/Inspiron 8xxx series used a 56 watt hour battery, while our test unit uses the new 66 watt hour battery which will be found in C840 and 8200.
This is a front view of our Dell test unit with both the battery and the C-Bay device removed from the front of the unit.
Our test unit weighs about 7.5 lbs or more, depending on the configuration. Like the Latitude C810 and Inspiron 8100, this means that in most configurations, you are looking at close to 8 lbs. The unit measures 1.75″(h) x 13.03″(w) x 10.87″(d). Again, this is the same as the C810/8100. Still, for a desktop replacement, this weight will be considered about the norm and far from the 8.5 lbs to 9 lbs for other desktop replacement units that we have seen.
The keyboard of our test unit is what we would consider to be good, but not as outstanding as the keyboard of the IBM A31p that we reviewed early this month. The keyboard action is firm, but a little on the stiff side. Over repeated use it might improve, but we saw no indication of this during our testing. The test unit features for the Alps StickPointer style pointing device as well as the Alps GlidePoint have become very popular among many users. The test unit features four pointing device buttons, two above the Alps GlidePoint and two below. We found the buttons to be of good size and the click action to be good. Unlike the Inspiron product line that allows you to customize the palm rests with different colors, our test unit was closer to the Latitude which offers no such option. This is to be expected, because the Latitude is all business. We found the palm rest area of the unit to be of reasonable size, but it did not have the feel that we liked so much on the A31p. Of course, this is a matter of sheer preference, and the size of your hands will have a lot to do with your comfort level.
Breaking In The Lid
Under the keyboard is a heat pipe to help remove heat from the GeForce4 Go GPU. We think this is an innovative solution that does help reduce the temperature of the GeForce4 Go GPU. During our use of the system, we didn’t find the keyboard getting hot, so it seems to work well. We think this was an excellent and innovative idea by Dell.
In this picture you can see the heat pipe that Dell uses to remove heat from the GeForce4 Go GPU. The well designed innovation didn’t cause the keyboard to get hot at all, which was a big surprise to us.
The test unit features a 15″ UXGA active matrix TFT display, with a native resolution of 1600×1200 with 16 million colors. As you already know from the title of this article, the unit can be configured at the point of sale with the GeForce4 Go in either the 64MB or 32MB configuration. We would rate the display quality as good, but there are a few things about it that we don’t like. Ever since the early Latitude C800 and Inspiron 8000, we have never been too crazy about the way that the display attaches to the machine.
Our Dell test unit’s 15″ UXGA active matrix TFT display from another angle.
The hinge quality was somewhat sloppy on our review unit, which caused the display to move around a little bit more than usual. The display field of view is more restrictive than that of the IBM A31p, which can be both good and bad, as we pointed out in our A31p review. We noticed that as we changed our angle of view, the display had a tendency to get a little lighter or darker than normal. In some ways, I think we were spoiled by the FlexView technology that IBM A31p offered. We do think that the overall quality of the display is good, but, in our opinion, Dell needs to try do something to stiffen up and strengthen the hinges, which we think is one of the continued weak points of the Dell Latitude 8xx and Inspiron 8xxx product families. The lid of the unit is of the same durable plastic that Dell has been using in these product families. We found the lid to flex a little more that we would like and we would have preferred it to be a bit stiffer to help avoid putting pressure on the back of the panel, which can cause problems. The latch on the lid also takes a little practice to master, but we noticed that, as it was broken in over time, it became a little easier to use.
A view from of the top of our test unit with the lid closed.
Rigged For Networking
Our review unit was wired for 802.11b using the MiniPCI slot, however, we were not provided with a MiniPCI 802.11b wireless card for our testing, so we cannot really give you any idea of how well the internal built-in antenna might perform.
The front of our test unit is where the C-Family Media Device Bay is located. Dell offers a variety of devices that can be placed into the C-Bay, including: 250MB Iomega Zip drive; 2nd Hard Drive Bay; 24X/10X CD-ROM; 8X DVD; 8X4X24X CDRW; CDRW/DVD, which is 8X/8X/8X/24X; and, of course, it can also hold a second battery to double your battery life. The C-Bay in our test unit was compatible with C-Family Media Device family members, so if your IT shop already has a large stock of C-Family Media Devices, this is good news for you.
Next to the C-Bay, you will find a space for the 66 watt hour battery that is required by our test unit. Like past battery designs from Dell, the battery has the same form factor as previous Latitude battery designs. The battery includes a battery test indicator which has five lights to indicate the current charge level.
On the left side of our test unit is the Dell fixed optical bay. This bay can be pre-configured with a variety of options from Dell including: 24X/10X CD-ROM drive, which is standard; optional 8X DVD; optional 8X4X24X CDRW; and even the optional CDRW/DVD, which is 8X/8X/8X/24X. (Dell is also offering a CDRW/DVD 16X/10X/8X/24X drive that can be installed in select units.) This device is bolted to the unit, so although it could be exchanged, it is not recommended by Dell.
As you can see in this picture, it is possible to remove the fixed optical drive from the fixed optical bay in our test unit, but it isn’t something that Dell would suggest. It uses a standard notebook style optical device, however.
In addition, on the left side you will find an integrated PC-Tel Modem 2304WT V.92 software modem. For network connectivity, we found the integrated 3Com 3C920 10/100 Fast Ethernet based NIC. Also, you will find the S-Video Out, which can be converted to RCA style composite out with the use of the supplied converter cable. The unit also has a connection for the standard locking Kennisington style devices. You will find a small speaker on both the left and the right side of the unit, to provide a full and rich sound.
What Is A Final Engineering Sample?
On the right side of our test unit, you will find the removable hard drive tray. The tray on our test unit can handle up to a 12.5mm hard drive. Our unit was equipped with the 60GB IBM 5400RPM Travelstar. As we have talked about in some of our previous reviews, hard drive performance is still a factor, but the Dell comes with the fastest hard drive currently available. We found the performance of the 60GB Travelstar to still be pretty good when compared to some of the other hard drive solutions for notebooks that are currently available, but they all still lag behind the performance of your typical desktop system.
Dell has packed a lot into the right side of the unit. We removed the 60GB 12.5mm hard drive from the unit, so that you could see how it slides back into our test unit.
Below the hard drive, you will find an Infrared port. Next to this, you will find two 32-bit Cardbus slots in a stacked configuration. Below these, you have 1394 Firewire port, Mic In, Head Phone Out, and Line Out. The Line Out can be reconfigured to provide digital out services via an option within the control panel. Also, you see one vent port that is used to help vent the heat from the processor.
The rear of our test unit has the following ports: Power In; PS/2 Keyboard/Mouse; VGA Out; C-Dock; LPT Printer; COM (Serial); and two USB ports. The C-Dock port is compatible with the C-Dock or C-Port Replicator devices that you already own. Of course, this provides additional value to those IT shops that are already using Dell notebooks within their enterprise.
The back of our test unit looks like the back of most modern notebooks in this class. The assortment of ports is standard. The two fan configuration on the unit is a little bit out of the ordinary, but serves it well.
What Is A Final Engineering Sample?, Continued
Dell has chosen to go with a small, two fan configuration to remove heat from the heat pipe out of the rear and left corner of the notebook. At times, this back part of the notebook can reach some high temperatures that are a little more than warm to the touch. The two fans in the back of the unit do not run all of the time. In fact, at times only one of them might come on for a minute or two, depending on the temperature and how much heat needs to be dissipated from the unit. When the unit gets really hot, both fans might come on for an extended period of time. With both fans running, it can get a little noisy, but we have noticed some improvement in our test system over other Dells that we have tested. Overall, the fans don’t stay on too long, even when running the most intense of benchmarks, but with both of them ramped up to full speed to cool the system, you will definetely hear that they are on.
In our test unit, Dell uses a heat pipe configuration along with the two fans to keep the unit cool.
The bottom of the unit does not present much of a change from other Dell Latitude 8xx/Inspiron 8xxx models. It comes complete with one MiniPCI slot that is standard. On our unit it was not populated, but as you can tell from the picture, it had an antenna hook-up for 802.11b wireless. In addition, it offers two memory slots, which can be populated up with PC2100/266MHz DDR to reach it’s maximum memory limit of 1GB of RAM.
You really don’t see much on the bottom of our test unit other than the MiniPCI slot and the two memory expansion slots.
From what we could see from our test unit, Dell really didn’t introduce anything new. Instead they took a unit that was already successful and upgraded it to the latest processor, chipset, and the latest graphics technology, sticking with a proven winner. While we would have liked for Dell to show us something new, our test unit didn’t disappoint, but we sure would have liked them to show us some new and innovative features rather than just sticking with the status quo.
Test Setup
Dell Final Engineering Sample Technology To Appear In Dell C840 & C8200 |
|
Processor | Intel Mobile Pentium 4m 1.7GHz |
Motherboard | Based On The Intel i845 Mobile Chipset |
Memory | 512MB PC2100 DDR SDRAM (as Tested) – Min 128MB / 1024MB MAX |
Display | 15″ UXGA 1600×1200 |
Graphics Card | NVIDIA GeForce4 Go 64MB DDR / 220MHz Core/440MHz Memory |
Modem | PC-Tel Software Modem 2304WT V.92 |
Ethernet | 3Com 10/100 3C920 Intergrated Chipset |
Battery | Li-Ion Batter Life – 2 Hours – 66 watt hour |
AC Power Adapter | 90 Watt (20V 4.5A) |
Spindle/Expansion Bays | 1 – Fixed Optical Bay 1 – C-Family Media Device Bay |
Ports | 2 USB – 1 1394/Firewire – LPT Printer – COM Serial – Svideo Out – Infrared – C-Dock/C-Port Replicator – VGA Out – PS/2 Keyboard/Mouse – Audio Mic/Headphone/Line Out – AC Adapter In – 2 Cardbus Card Slots (Stacked Configuration) |
Dimensions / Weight | 1.75″(h) x 13.03″(w) x 10.87″(d) – 7.5lbs or more depending on the configuration |
BIOS Version | 0.0 |
Test OS | Microsoft Windows XP Professional – All Current Patches Applied |
Pre-Load OS | Microsoft Windows XP Home/Professional – Windows 2000 SP 2 |
OS Support | Microsoft Windows XP & Windows 2000 Service Pack 2 Only |
Specifications Of The A31p Model H6U | |
Processor | Intel Pentium 4 Mobile – 1.7Ghz w/512K Cache |
Motherboard | Motherboard Based On The Intel i845 Mobile DDR Chipset |
Memory | 512MB PC2100 DDR SDRAM |
Display | 15″ UXGA FlexView 1600×1200 |
Hard Drive | 60GB ATA-100 12.5mm 5400 RPM Hard Drive |
Graphics Card | ATI Mobility Fire GL 7800 128-bit 64MB DDR – 4X AGP |
Network | 802.11b Wireless – Actiontec |
Modem/Ethernet | 56K Softmodem by Agere Systems 10/100 Intergrated Ethernet by Intel Bluetooth Connectivity By Actiontec |
Battery | Li-Ion Batter Life – 2 Hours |
Ultrabay | Ultrabay Plus Ultrabay 2000 |
Ports | 2 USB Ports 1394 Firewire S-Video Out Infrared Dock/port eplicator Connector VGA Out Audio – Headphone/line out, Mic In, ine IN AC Adapter In IBM Embedded Security Systems 2 PC-Card/Cardbus Slots – Stacked Configuration |
Dimensions – Weight | 13″x10.7″x1.8″ – 33.02cm x 27.2cm x 4.6cm 7.7lbs – 3.5kg |
Warranty | Limited 3 Year |
Test OS | Microsoft Windows XP Professional w/all current updates applied |
Preloaded OS | Windows XP Home/Professional / Windows 2000 |
Supported Operating Systems | Windows XP Home/Professional – Windows 2000 – Windows ME – Windows 98SE – Windows NT 4.0 (All devices not supported w/NT 4.0) |
Included Software | Access ThinkPad, Access Connections, Access Support, Adobe Acrobat Reader, ConfigSafe (Windows 2000 models Only), MGI VideoWave (A31p Models Only), PC-Doctor, ThinkPad Utilities, Update Connector, IBM RecordNOW by Veritas (CDRW Models Only), Client Security Software (Embedded Security Subsystem Models Only – Download), IBM Data Transfer, IBM Rapid Restore, IBM Director Agent, Lotus SmartSuite Millennium, and Lotus Notes Stand Alone Client |
Testing Configuration – Dell Inspiron 8100 | |
Processor | Intel Pentium IIIm 1.2Ghz |
Display | 15″ UltraXGA+ TFT Display |
Memory | 256MB PC-133 SDRAM |
Hard Drive | 20GB ATA Hard Drive |
CD Drive | 8X DVD-ROM Drive |
Floppy | 1.44 Floppy Drive |
Network | MiniPCI 10/100 NIC / 56K Modem Combo |
OS | Microsoft Windows XP Professional – All Current Updates Applied |
Graphics Card | ATI Mobility Radeon 7500 64M – Using Driver Version 7.64RC4 – A01C NVIDIA GeForce2 Go 32MB DDR – Using Driver Version 16.40 – A07 |
Benchmarks
We wanted to use our Dell test unit in a one-on-one showdown between the NVIDIA GeForce4 Go and the ATI Mobility Radeon 7500. With Dell’s continued use of the mobile graphics subsystem, and being the optimists that we are, we attempted to install the ATI Radeon Mobility 7500 into our test unit. However, due to some compatibility issues, we were not able to present you with what would have been a better look at the two current mobile graphics technologies. With the ATI Mobility Radeon 7500 installed in our test system, although we could get into Windows XP, once the drivers were loaded, no matter which version we used, we could not get the Mobility Radeon working in our test unit. We believe that this was due to a BIOS/Video BIOS compatibility issue. (So, those of you already wondering if you are going to be able to get the GeForce4 Go 64MB working in your Dell Inspiron 8100 or Dell Latitude C810, that answer is, we would doubt it!) Hopefully, we will get the chance to pit these two titans on an equal platform in the future.
In the meantime, we planned to have a look at the performance scores of both the IBM’s A31p and Dell’s Inspiron 8100 in relation to our test unit. We think that this will give you a good look at the performance difference between the three units.
To give you additional insight into PowerMizer’s impact on system performance, we ran most of the tests in all three PowerMizer modes. This should give you some additional help in determining the overall impact of using PowerMizer. From the numbers, we found that PowerMizer can be useful in many situations. Again, we should note that Dell is not planning to support or offer PowerMizer at this time.
It is worth pointing out that at this point there are some very good deals to be had on the older Pentium IIIm technology, and many have asked us if we think there is still value in purchasing these systems. I guess that our best answer would be that, “Yes, depending on what you use the system for, you might be able to get by just fine with the purchase of the current Pentium IIIm systems.” Not everyone needs to have the latest in technology and the flexibility of a more robust mobile graphics subsystem, so we would encourage you to also explore some of the other Pentium 4m notebooks that are going to be released shortly with both the ATI Mobility Radeon and the NVIDIA Geforce2Go graphics options. A good example of this is the Winbook n4, which is based on the Pentium 4m, but uses the less expensive GeForce2Go solution, leading to a lower overall cost. As an example, the Winbook n4 will sell for close to what you can purchase the Dell Latitude C810, Dell Inspiron 8100, and the Gateway Solo 9550 series for; of course, this will depend on the configuration. Comparison shopping is always your key to getting the best overall deal, but if you are an avid reader of THG, you already know that.
As we have explained in our past notebook articles, platform differences do affect overall performance. These performance benchmarks should be viewed not as a direct performance comparison between the three notebook platforms, but as a performance comparison of processing power of both CPU and GPU more than anything else. Although some conclusions can be drawn by looking at the numbers, by no means does everything become clear, that is for sure.
Since the unit provided to us was a Final Engineering Sample, we don’t expect the numbers to be that different from the final shipping product. We do plan to run all of the tests again on either the Dell Latitude C840 or Inspiron 8200 when one of those units is made availiable to us. Once again, because our test unit is a hybrid and uses technology from both what will be the C840 and the 8200, no direct comparison to these units is possible because this unit doesn’t fully match either configuration availiable to the consumer.
Quake 3
As you can see here, at the 1024x768x32 resolution, our test unit equipped with the GeForce4 Go is clearly faster than the A31p equipped with the ATI Mobility Radeon 7500. The performance gap of 23 frames is a significant margin of victory. As for PowerMizer’s performance, in the balanced mode it is still very playable, but in battery saving mode the FPS results are almost half those of the balanced mode.
Testing Quake III at 1600x1200x32 showed a little bit of a different story. Our test unit and the A31p were dead even. No matter how many times we ran the tests, the results were always 1 frame +/- . So, at the higher resolutions, it is more of a toss up on who the winner really is. It is worth noting that that the Inspiron 8100 with the Mobility Radeon 7500 doesn’t cut such a bad number with a score of 55.5, which is still within 8.8 frames of the Pentium 4m platform systems. The PowerMizer numbers at this resolution are less impressive. At 41.6 FPS in the balanced mode it would still be playable, but when the number shrinks to 21.3 FPS in the battery saving mode, it is no longer worth it. We would not suggest the use of PowerMizer for a game of this type at this resolution.
3DMark 2001
The 3D Mark 2001 tests showed an interesting story, while it is evident that the GeForce4 Go does have a bit of an edge on this test. The test unit was able to beat the A31p by a margin of 377, which indicates that it does have a small performance edge. The Dell Inspiron 8100 trails the pack by a margin of almost 1000, which is significant. We see about the same results for PowerMizer again in these tests. It is obvious that if you are using something that requires GPU processing power, then using PowerMizer really cuts down the performance.
Lightwave and the Skull_Head_Newest.LWS
We have used Lightwave and the Skull_Head_Newest.LWS test for some time. We thought it might be a good idea to try to get an estimate of the processing power of the Pentium 4m between the test system and the A31p. We felt that this intensive test might give us an overall indication of how much processing performance there really is between the three platforms. Here, the test unit and the A31p were about equal, which is what we would have expected on this test. The edge here went to the A31p, but, again, the results of this test were almost too close to call. The PowerMizer results told another story altogether on this test. If these are the kind of applications that you are running on your notebook, for a small performance hit, using PowerMizer can help extend your battery life.
SPECViewPerf
We wanted to get a feel for the performance difference in optimizations between the drivers for the test system and the A31p. Of course, the A31p features an ISV certified driver, which should give an edge on this test, and it did. Of course, when NVIDIA releases the Quadro4Go, this test will have more meaning. In the meantime, as we showed you in our testing of the A31p, it would appear that a more balanced approach to optimizing the driver has been taken by ATI. For this reason, if you run a lot of apps that are targeted for this kind of performance, looking at the Radeon Mobility 7500 might make a little more sense. On the other hand, the GeForce4 Go isn’t optimized to run these kind of tests and leaves this stuff to the Quadro4Go. PowerMizer didn’t seem to make much of a difference with this test either. You can gain some additional power conservation in the max battery life mode for a very small performance hit.
BatteryMark
We took this chance to use the new BWS BatteryMark 1.0 to see what kind of simulated battery performance we could expect in each of the three PowerMizer settings. BWS BatteryMark replaces the older BatteryMark 4.0, which was no longer supported as of August of last year. The new BWS BatteryMark uses Business Winstone in its testing script in an attempt to obtain a more realistic projection of expected battery life.
Due to a compatibility problem with Windows XP and the new BWS BatteryMark 1.0 that we were unable to pinpoint, we had to run the BWS BatteryMark 1.0 tests under Windows 2000 Service Pack 2. We reloaded the system with Windows 2000 Service Pack 2 and applied all of the current updates. Of course, as with all BatteryMark scores, these should only be taken as a best-case guideline for battery performance, your results may vary.
As we see here, the performance of BWS BatteryMark 1.0 across all three PowerMizer settings was the same. Although from our testing it was clear that the GeForce4 Go was drawing less current, it was also clear that PowerMizer does not have much of an impact on the BatteryMark test. This is due in part to the way that the BatteryMark test is designed. PowerMizer can only improve battery life, but we have to question how noticable this will be to the consumer.
Real World Battery Life
In our DVD playback testing, we used “The Patriot” as our test DVD. Many things can affect battery performance. We prefer to think of these numbers as an informal test rather than something more that that. We know from your feedback that this continues to be an important consideration in the purchase of many laptops, but many things can effect the outcome of this test. The time of 1 hour and 53 minutes that was turned in by our test system was still a little short of what we expected.
Currently, if you want the maximum DVD playback time, at least for right now, you are better off sticking with a Pentium IIIm based system. We would have expected PowerMizer to make more of a difference in this test as well, but after speaking with NVIDIA we learned that during their studies they have found that users were more interested in quality than battery life. We find this quite puzzling beause most of our readers tell us the exact opposite — which is to say, that they would rather have additional battery life. The fact that NVIDIA is focused on playback quality does answer the question of why performance wasn’t better, but we still would have liked to have the option to try to increase battery life.
At this point there is no doubt that ATI does have an edge in DVD playback battery life performance. We would like to see NVIDIA focus on improving the length of playback time, because otherwise we are going to have to learn to only watch movies that are 1:45 long, or make sure that we have a second battery in our C-Bay, which is only going to make our shoulders more sore. Given the performance of the unit in our BWS BatteryMark 1.0 test, could any additional optimizations or improvements be made to current DVD-ROM drives in notebooks so that they would use less power? This might be one key in improving notebook performace on this test.
Sysmark 2002
The results from the Sysmark 2002 Overall score give a slight edge to our test system. However, it should be noted that the Inspiron 8100 isn’t that far behind. If your primary target is productivity applications, Pentium IIIm systems are still worth considering.
In the Office application productivity testing, our test system was able to best the others, which should come as no surprise as it was able to win the overall score, as well.
Conclusion
The GeForce4 Go does take the performance crown in the majority of our testing, but the Mobility Radeon 7500 (and Mobility FireGL 7800) wasn’t far behind. When the GeForce4 Go is coupled with our test system, this makes quite a combination. Of course, the addition of the GeForce4 Go in the 64MB configuration adds an additional $99 to the cost. Still, when you are looking at system of this caliber, it is understood that, to have the best features, you can expect to pay a little more.
Our Dell test system is what we have come to expect in the Latitude 8xx/Inspiron 8xxx product line. Very few changes in the product design gives Dell the ability to deliver a product that is worthy of consideration for purchase. Cross platform peripheral compatibility across the product line is an advantage for IT shops that have already standardized on Dell notebooks. This value should not be overlooked, as it can help your company continue to save money by recycling old peripherals for use with newer systems.
We expect the pricing for our test system to start at a very reasonable $2700, but then escalate to about $3300 for our tested configuration. This price is still acceptable, though, when compared to other systems in the same class. Again, this would only be a guess based on information on Dell’s web site.
The GeForce4 Go does deliver on its performance promise and does deserve purchase consideration, but the lack of notebook vendors that are using the GeForce4 Go will cause consumers to continue to pay a higher price for it. However, depending on your applications, you might be just as happy with the purchase of the Radeon Mobility 7500 equipped notebook.
As we can see from the results of our PowerMizer in-depth testing, there is no doubt that PowerMizer reduces the power requirements of the notebook, but again, we question how much real-world benefit users will see. The maximum battery life setting uses 25% less power, but again, we question how much real-world benefit the end user will see. 25% does sound like a lot, but it appears that it will take time for other system components to catch up with NVIDIA’s lead and reduce overall battery consumption. New advancements in the science of battery chemistry are also part of the puzzle in getting longer battery life, and of course this cannot be overlooked. At least for right now it would appear to us that ATI continues to have a lead over NVIDIA in power management performance. We will continue to take the OEMs to task for not making use of the full power management support provided by ATI and NVIDIA; more can be done to increase battery life.
For the time being, each of the new mobile GPUs has areas where they have gained ground. Selecting a clear winner would be difficult, as there seems to be a good balance between each of the cards, but performance of the GeForce4 Go would appear to have a bit of an edge over the Mobility Radeon 7500 at the moment. The Mobility Radeon 7500, on the other hand, does show better performance in a few areas where the GeForce4 Go is lacking. The bottom line is that, depending on what you are using your notebook for, the decision is almost a draw.
Until such a time that we are able to test optimized versions of both mobile video solutions on the same platform, in the same machine, the jury is still out. We would welcome an opportunity to do this, but currently it looks like the odds are against it happening anytime soon.