|
楼主 |
发表于 2008-9-21 20:25:14| 字数 14,147| - 中国–天津–天津 联通
|
显示全部楼层
Dell Latitude E6400用户评测 | (下)(原文)
Submitted by John Ratsey on Monday, September 08, 2008
From: NotebookReview
Performance
Note that results are for 2 x 1GB DDR2-800 RAM unless indicated otherwise. Furthermore, all the tests were carried out under Windows XP SP3. This can slightly affect the results. For example, tests on my Toshiba R500 revealed that wPrime gives a faster time under Vista than under XP.
The 2.4GHz Intel P8600 Core 2 Duo processor in my E6400 configuration sits in the middle of the range these days. I've had no issues with the performance of the 2GHz T7300 in my Zepto but a small boost in speed is always welcome, particularly if it comes with reduced power consumption. The Intel GM45 chipset is claimed to provide significant performance improvements compared with its predecessor as well as providing better power efficiency.
CPU and Memory
CPU-Z's report for the memory is below (CPU pages for both versions 1.40.5 and 1.47 since they see things slightly differently). I noted that the supplied 1GB of RAM was CL 5 but only running at CL 6. However, when I added a second PC 6400 RAM module with CL 5 timing then the memory ran at CL 5.
SiSoftware Sandra includes various benchmarking modules with a database of results. I used it to test the CPU performance. As I expected, there was limited difference between the P8400, P8600 and P9500 CPUs. There was a 20% improvement in the Dyrystone score compared to the T7300 in my Zepto (ie in proportion to the increased clock speed) but a 36% improvement in the Whetstone score, indicating improvement to floating point performance as well as the faster clock speed.
There has been discussion of the potential performance improvements offered by DDR3 RAM which is an option for the chipset. However, Dell decided to keep with DDR2 RAM. I have checked the Memory Bandwidth using SiSoftware Sandra. It is noticeable that the newer chipset has about 20% higher memory bandwidth than the Intel 965 mobile chipset. A symmetrical memory configuration gives the best bandwidth (I observed 5.52GB/s with 2 x 1GB) with a slight reduction for an asymmetrical configuration and a further reduction when only one memory module is used. Comparative results are needed for DDR3 RAM to see what performance improvement it provides over DDR2.
SuperPi
SuperPi is a single threaded benchmark which used to be popular as a measure for raw CPU performance. I include the result for historical purposes since it shows how the performance of a single core has improved over the past few years. The 2.4GHz P8600 CPU needed 49 seconds to complete the calculation to 2 million digits. This is 10 seconds faster than the 2GHz T7300 CPU and one third of the time needed by a 1.6GHz Pentium M CPU.
Comparison of SuperPi Results
Notebook
Time
Dell E6400 (2.4GHz Intel P8600 / Intel GM45 / 2GB RAM PC6400) 0m 49s
Zepto 6024W (2.0GHz Core 2 Duo T7300 with 800MHz FSB and 667MHz RAM) 0m 59s
Dell Latitude D830 (2.2GHz Core 2 Duo T7500, 800MHz FSB, 667MHz RAM) 0m 53s
Samsung Q70 (2.0GHz Core 2 Duo T7300 with 800MHz FSB and 667MHz RAM) 0m 57s
Dell XPS M1330 (2.0GHz Intel Core 2 Duo T7300) 0m 58s
Lenovo ThinkPad T61 (2.00GHz Core 2 Duo Intel T7300) 0m 59s
Samsung X60plus (2.0GHz Core 2 Duo T7200 with 667MHz FSB & memory speed) 1m 02s
Samsung Q35 (1.83MHz T5600 with 667MHz FSB and 533MHz RAM) 1m 16s
wPrime
I have been suggested that SuperPi should be superseded by wPrime which is multi-threaded. The 2.4GHz P8600 CPU completed the 32M calculation in 32.53s. This 77% of the time needed by the 2GHz T7300 CPU.
Comparison of wPrime Results
Notebook / CPU
wPrime 32M time
Dell E6400 (2.4GHz Intel P8600 / Intel GM45 / 2GB RAM PC6400) 32.53s
Lenovo T400 (Intel Core 2 Duo T9600 @ 2.8GHz) 27.410s
Zepto 6024W (Core 2 Duo T7300 @ 2GHz) 42.385s
Lenovo T61 (Core 2 Duo T7500) 37.705s
Alienware M5750 (Core 2 Duo T7600 @ 2.33GHz) 38.327s
HP Pavilion dv5t (Core 2 Duo P8400 @ 2.26GHz) 34.52s
HP Pavilion dv5z (Turion X2 Ultra ZM-80 @ 2.1GHz) 39.745s
Acer Travelmate 8204WLMi (Core Duo T2500 @ 2.0GHz) 42.947s
Zepto Znote 6224W (Core 2 Duo T7300 @ 2.0GHz) 45.788s
PCMark05
PCMark05 tests various aspects of a computer's performance in order to generate an overall score. With 2 x 1GB of RAM my E6400 scored 4,357 PCMarks. This is only a 7% improvement on my Zepto notebook. Investigation of this relatively poor showing merits further investigation when I have time and inclination. An asymmetric RAM configuration causes a slight reduction in the PCMark scores.
Comparison of PCMark05 Results
Notebook
PCMark05 Score
Dell E6400 (2.4GHz Intel P8600 / Intel GM45 / 2GB RAM PC6400) 4,357 PCMarks
Dell E6400 (2.4GHz Intel P8600 / Intel GM45 / 2GB RAM PC5300) 4,357 PCMarks
Dell E6400 (2.4GHz Intel P8600 / Intel GM45 / 3GB RAM PC6400) 4,212 PCMarks
Dell E6400 (2.4GHz Intel P8600 / Intel GM45 / 1GB RAM PC6400) 4,230 PCMarks
Lenovo T400 (2.80GHz Intel T9600, ATI Radeon 3470 256MB) 6,589 PCMarks
Lenovo T500 (2.80GHz Intel T9600, Intel X4500) 5,689 PCMarks
Zepto 6024W (2.0GHz Core 2 Duo T7300 and Intel X3100 GPU) 4,063 PCMarks
Zepto 6625WD (2.4GHz T7700, Nvidia Geforce 8600M GT 512MB) 5,123 3DMarks
HP Pavilion dv5z (2.1GHz Turion X2 Ultra ZM-80, ATI Radeon HD 3200) 3,994 PCMarks
Dell XPS M1330 (2.0GHz Core 2 Duo T7300, 8400M GS) 4,571 PCMarks
Samsung X60plus (2.0GHz Core 2 Duo T7200, ATI X1700) 4,555 PCMarks
Samsung Q70 (2.0GHz Core 2 Duo T7300 and 8400M G GPU) 4.491 PCMarks
3DMark05
With 2GB of DDR2-800 RAM my E6400 managed a score of 1503 3DMarks for 3DMark05. The test was run at 1024 x 768 resolution with no anti-aliasing. This result is over 50% better than I had observed with the T7300 CPU and Intel 965GM graphics. I have also included results for different RAM configurations. 2GB of DDR2-667 RAM gives a slightly better result than 3GB of DDR2-800 RAM while 1 x 1GB RAM causes a noticeable drop in performance which is probably caused by either a reduced GPU memory allocation and/or the slower memory performance in single channel mode.
Comparison of 3DMark05 Results
Notebook
3DMark05 Score
Dell E6400 (2.4GHz Intel P8600 / Intel GM45 / 2GB RAM PC6400) 1,503 3DMarks
Dell E6400 (2.4GHz Intel P8600 / Intel GM45 / 2GB RAM PC5300) 1,421 3DMarks
Dell E6400 (2.4GHz Intel P8600 / Intel GM45 / 3GB RAM PC6400) 1,414 3DMarks
Dell E6400 (2.4GHz Intel P8600 / Intel GM45 / 1GB RAM PC6400) 1,295 3DMarks
Zepto 6024W (2.0GHz Core 2 Duo T7300 and Intel X3100 GPU) 910 PCMarks
HP Pavilion dv5z (2.1GHz Turion X2 Ultra ZM-80, ATI Radeon HD 3200) 2,969 3DMarks
Zepto 6625WD (2.4GHz T7700, Nvidia Geforce 8600M GT 512MB) 6,047 3DMarks
Samsung X60plus (2.0GHz Core 2 Duo T7200, ATI X1700 256MB) 4,150 3DMarks
Dell XPS M1330 (2.0GHz Core 2 Duo T7300, NVIDIA 8400M GS) 3,079 3DMarks
3DMark06
With 2GB of DDR2-800 RAM my E6400 managed a score of 944 3DMarks for 3DMark05. The test was run at 1024 x 768 resolution with no anti-aliasing. This result is nearly 70% better than I had observed with the T7300 CPU and Intel 965GM graphics. I have also included results for different RAM configurations. 2GB of DDR2-667 RAM gives a slightly better result than 3GB of DDR2-800 RAM while 1 x 1GB RAM causes a noticeable drop in performance which is probably caused by either a reduced GPU memory allocation and/or the slower memory performance in single channel mode.
Comparison of 3DMark06 Results
Notebook
3DMark06 Score
Dell E6400 (2.4GHz Intel P8600 / Intel GM45 / 2GB RAM PC6400) 944 3DMarks
Dell E6400 (2.4GHz Intel P8600 / Intel GM45 / 2GB RAM PC5300) 893 3DMarks
Dell E6400 (2.4GHz Intel P8600 / Intel GM45 / 3GB RAM PC6400) 873 3DMarks
Dell E6400 (2.4GHz Intel P8600 / Intel GM45 / 1GB RAM PC6400) 786 3DMarks
Zepto 6024W (2.0GHz Core 2 Duo T7300 and Intel X3100 GPU) 561 3DMarks
Zepto 6625WD (2.4GHz T7700, Nvidia Geforce 8600M GT 512MB) 3,017 3DMarks
Lenovo T400 (2.80GHz Intel T9600, Intel X4500) 809 3DMarks
Lenovo T400 (2.80GHz Intel T9600, ATI Radeon 3470 256MB GDDR3) 2,575 3DMarks
Fujitsu Siemens Amilo Xi 1526 (1.66 Core Duo, nVidia 7600Go 256 MB) 2,144 3DMarks
Cinebench
Cinebench is a good rendering benchmark tool based on the powerful 3D software, CINEMA 4D. Its rendering tasks can stress up to sixteen multiprocessors on the same computer. It is a free benchmarking tool, and can be found at http://www.cinebench.com. It has been recently updated from version 9.5 to 10 and I have included some results below for both versions. Cinebench also includes an OpenGL benchmark which will be of interest to those people who use software which uses OpenGL. While basic rendering performance has increased approximately in proportion to the CPU speed, the OpenGL performance has increased by 50% to 60% relative to my Zepto 6024W.
HD Video Playback
People have asked about the HD Video playback performance. I downloaded and played the Terminator 2 trailers from Microsoft. The E6400 played these back without hesitation. The fan came on, but only at the slow speed. For comparison I played the same samples on my Zepto, which seemed to miss a few frames of the 1080p sample. This tends to confirm that Intel have included HD video decoding in the GM45 chipset.
I used RMClock to provide insight into the CPU behaviour. The upper set of graphs is for the E6400 and the lower set is for the Zepto 6024W (T7300 + 965GM). The most revealing part of the graphs is the third row (CPU voltage and frequency). The P8600 was able to handle some of the playback at the SLFM frequency (800MHz) but speeding up to 1.6GHz at times (more on the 1080p sample). In comparison, the T7300 / 965GM combination required the CPU to run at up to its maximum speed. (Note for anyone who looks at the temperature: RMClock seems to be reading the P8600 temperatures 5°C low).
CPU Utilisation during HD playback on E6400 (left = 720p, right = 1080p)
CPU Utilisation during HD playback on Zepto 6024W (left = 720p, right = 1080p)
Power Systems
Dell's information about the E series included description of a new slim 90W PSU. However, the PSU supplied with my E6400 was a D series 90W "brick" which looks even bigger than the PSU for my Zepto. I had also ordered the optional slim 65W auto / air / AC adaptor since 65W is plenty enough power for a notebook with integrated graphics. It is often convenient to have one PSU at home and another at the work place. I am very impressed by the size and convenience of the slim 65W PSU. However, I am annoyed that the BIOS detects it and then reduces the computer's performance. See Annex 2 for more details.
The standard and slim PSU with a cell phone for scale.
I have a mains power meter and the maximum power drain when running a benchmark program such as 3DMark06 is less than 40W at the mains socket (so maybe 35W at the computer). Under light usage the mains power drain is around 14 to 18W. The only time I have seen a power drain higher than 40W is during battery charge. The standard charging rate is about 35W (reducing as the battery charge approaches 100%) so full load and full charge could combine to draw about 70W. Most notebooks reduce the charge rate if the combined power requirement exceeds the PSU capacity. There is also an express charge option which has to be enabled in the BIOS. Dell claim this will charge the battery to 80% capacity within one hour.
Heat and Fan Noise
In my experience, the E6400 raises the bar on the heat and noise management. I initially though the fan wasn't working because it was so quiet. In fact, the metal heatsink and chassis absorbs a lot of heat so the fan has little work to do when the computer is under light load. There appears to be three fan speeds: slow, fast and off. The lowest fan speed is unobtrusive but the fan does get audible at the fast speed but needs a couple of minutes of stressed CPU to trigger it. My configuration probably benefits by having one of the lower power rated CPUs and the integrated graphics combined with a cooling system designed for more powerful configurations. The maximum CPU temperature I have observed was 75°C just before the fan switched to fast speed during a stress test.
After a week of watching the fan behaviour I am wondering whether the fan operating rules are designed to minimise the thermal cycling of an nVidia GPU instead of providing the most pleasant experience for users with the Intel GPU. On many notebooks the fan responds very quickly to changes in CPU activity. On this computer it is very noticeable that once the fan does switch to the fast speed then it stays running at fast speed for several minutes after the CPU activity has stopped. A similar elapsed time rule also appears to apply for the slow fan speed. This is illustrated by the MobileMeter plot below (each grid line on the horizontal scale represents about 5 minutes). The rising curves indicate the fan is off and the temperature gradually rises until the fan comes on. The temperature then drops quite quickly but the fan keeps running for 10 minutes or more with the cooling system only slightly above ambient temperature before the fan finally stops and the temperature starts to climb again. I hope Dell will refine the BIOS to take account of the GPU type. The HDD stays very cool - normally less than 40°C.
Idle temperature variation
Battery Life
Dell has heavily advertised the 19 hour operation capability of the E6400. That is with a 9 cell main battery and the 12 (smaller) cell battery slice, but the inference is that a 6 cell battery should be good for 6 hours operation. This requires an average 9W power drain for a healthy battery. My conclusion is that 6 hours, maybe a little longer, is feasible but requires economical operation such as a dim backlight, no wireless and no applications which provide significant CPU usage. Any unused hardware needs to be disabled. However, 5 hours operation is easily achievable with the 6 cell battery. Sitting outside in the sunshine with the display very bright and the wireless on indicated a power drain suggesting 4 hours of operation. However, the battery will progressively wear and the run time will reduce. My fully charged capacity has dropped 1.5% in the past week. Versions of the E6400 with the T series CPUs, the nVidia GPU or faster hard disks can be expected to give shorter battery run times.
Battery capacity on arrival and one week later
The Dell Control Point software includes power configuration options aimed at reducing power drain when running on battery such as turning off unneeded devices. Power saving measures include putting the display into 16 bit colour and reducing the refresh frequency. However, while the software is quick to enable these power saving features when switching to battery I am not convinced it is as good at resetting everything to best performance when running on mains power. I have found I have had to manually restore the optimum display settings.
Some spot observations of the power drain show 8W to 11W under light usage with wireless on; DVD playback of 15W to 17W and power drain during a CPU stress test of 33W to 35W.
I have run my standard DVD playback test using "the 3 hour Dances with Wolves". However, I didn't quite get to the end for three reasons (i) I was not using the Dell optimised battery saving profile since it cut the DVD drive and the audio; (ii) I was working on this review while playing the DVD; and (iii) the last few percent of the battery charge seemed to disappear quickly. I am fairly confident that with tweaking of the power management the three hour barrier can be passed quite easily. I also noticed that the bundled Dell version of PowerDVD was playing the DVD continuously. Some versions of PowerDVD play the disc intermittently and use the RAM as cache, which is more power efficient (see my Toshiba R500 review).
If anyone is wondering how much power the backlight uses, I have made some observations of total power drain when running on battery. This shows a difference of nearly 4W between the minimum and maximum brightness. Fortunately, a brightness of a couple of steps above minimum is usable.
Conclusion
The pleasure of looking at the bright and clear LED backlit WXGA+ display offsets the displeasure of having a computer that is not as light as expected. All Dell's information indicates that this computer should be lighter than it actually is. So far, this computer has been a pleasure to use.
The build quality is excellent but Dell really should have found space for some better loudspeakers. The very smooth bottom will make it easier for the computer to launch itself towards the floor. Perhaps that is why Dell is keen on providing the drop-proof storage options.
Pros
Excellent display (LED backlit WXGA+ option)
Very rugged build
Good performance
Up to about 6 hours light usage with a 6 cell battery
Cool and quiet, except under heavy load when the fan runs fast
Easy access to most components
Cons
Overweight
Big standard power brick (what happened to the new E series slim 90W PSU?)
BIOS does not make proper use of the optional slim 65W PSU
Poor quality audio from the surprisingly small speakers hiding behind those big grilles
Overweight (yes, I said it twice)
(End)
[ Edited by standbyxiaoxu on 2008-9-21 21:00 ] |
|