The Card and PowerPlay 6.0
So, we have already mentioned that this part is similar to what will some day be the desktop X1600 series. Clock speeds and memory size aside, the part looks alright for its target market. The card needs to be able to run moderately sized panels at good quality and speed. Of course, battery life is important in this market as well, but we didn't have enough time with the system to run any battery life tests. The ASUS A7V in which we tested the MRX1600 was running a 2GHz Pentium M with 1GB of RAM. The system has a 1440x900 wide screen panel attached to it, which ends up requiring about the same pixel pushing power as a 1280x1024 display.
We ran a few brief tests just to see how the system performed. We wanted to know if we could play games at the maximum resolution without any problems. The A7V ran the tests that we tried quite well. Here's a look at our data.
PowerPlay 6 has most of the same features of PowerPlay 5. The GPU is able to run at 0.95V, the number of PCIe lanes is dynamically configurable, and clock gating is still a major feature, but there is one new addition that is quite interesting. The concept isn't new, but this is the first time that ATI has gone down this path.
We've spoken about leakage currents in some of our CPU articles before, but the basic problem of leakage is that high speed transistors end up drawing power even when they are off. One of the unfortunate side effects of trying to make it easier for electrons to flow through something quickly and easily is that it becomes more difficult to prevent them from doing so. And process shrinks don't help the problem either. ATI is combating the problem in its latest mobile parts by implementing reverse biasing. The idea is that in lower power modes, current flow through transistors that are off can be reduced if a voltage is actively applied in the opposite direction. ATI indicates that their method can reduce power by up to 20%.
So, we have already mentioned that this part is similar to what will some day be the desktop X1600 series. Clock speeds and memory size aside, the part looks alright for its target market. The card needs to be able to run moderately sized panels at good quality and speed. Of course, battery life is important in this market as well, but we didn't have enough time with the system to run any battery life tests. The ASUS A7V in which we tested the MRX1600 was running a 2GHz Pentium M with 1GB of RAM. The system has a 1440x900 wide screen panel attached to it, which ends up requiring about the same pixel pushing power as a 1280x1024 display.
We ran a few brief tests just to see how the system performed. We wanted to know if we could play games at the maximum resolution without any problems. The A7V ran the tests that we tried quite well. Here's a look at our data.
| Day of Defeat on Mobility Radeon X1600 | ||
| 1024x768 | 1440x900 | |
| No AA | 43.7 | 42.7 |
| 4xAA | 43.7 | 42.8 |
| Quake 4 on Mobility Radeon X1600 | |||
| 800x600 | 1024x768 | 1152x864 | |
| No AA | 60.9 | 45.7 | 39 |
| 4xAA | 40.8 | 30.8 | 26.6 |
PowerPlay 6 has most of the same features of PowerPlay 5. The GPU is able to run at 0.95V, the number of PCIe lanes is dynamically configurable, and clock gating is still a major feature, but there is one new addition that is quite interesting. The concept isn't new, but this is the first time that ATI has gone down this path.
We've spoken about leakage currents in some of our CPU articles before, but the basic problem of leakage is that high speed transistors end up drawing power even when they are off. One of the unfortunate side effects of trying to make it easier for electrons to flow through something quickly and easily is that it becomes more difficult to prevent them from doing so. And process shrinks don't help the problem either. ATI is combating the problem in its latest mobile parts by implementing reverse biasing. The idea is that in lower power modes, current flow through transistors that are off can be reduced if a voltage is actively applied in the opposite direction. ATI indicates that their method can reduce power by up to 20%.
Facebook
Twitter
RSS
30 Comments
View All Comments
Doormat - Monday, December 5, 2005 - link
I fully expect this chip to show up in Intels powerbooks that'll be annouced next month.acejj26 - Monday, December 5, 2005 - link
There's no way they fit the GPU die and all four of those memory modules on a package that is 46 mm^2. If it's a square, then that's less than 7 mm in each direction. That surely is a typo or mistakeDerekWilson - Tuesday, December 6, 2005 - link
Wow, very sorry ... that was a typo.I added that line to Josh's article. I appologize for not catching the mistake earlier.
42mm x 42mm is still very small.
Sunrise089 - Monday, December 5, 2005 - link
Maybe you could correctly quote the article and not insert a space that isn't there? It isn't Forty Six square mm, its Forty Six mm squared. Like 4 sides, each side 46mm long. This is like middle school stuff here.P.S. - I know this answer isn't nice, but above poster shouldn't call out the author before getting his/her facts strait.
Visual - Monday, December 5, 2005 - link
you're a doofus if you think a space means that much. both ways to write it mean the same to me, and surely it should be corrected.Sunrise089 - Monday, December 5, 2005 - link
Call me what you want, but I was capable of reading it correctly, apparently that is too difficult for others. I understand the reason people are consufed, but Square MM and MM Squared are two different commonly used concepts.tayhimself - Monday, December 5, 2005 - link
No theyre not. They are the same thing. Its a typo, in a horribly written article. No comparisons of battery life etc. Oh, and that is one hideous laptop.cryptonomicon - Tuesday, December 6, 2005 - link
my god it is hideousScrogneugneu - Monday, December 5, 2005 - link
46 Square mm refers to 46mm².46 mm squared refers to the length of one side of the chip, 46 mm, which still has to be squared, giving 21.16cm².
huges84 - Tuesday, December 6, 2005 - link
I disagree.I say that:
46mm^2 = 46 mm^2 = 2mm*23mm
and
(46mm)^2 = 46mm*46mm = 2116mm^2
Anytime you square a unit you are only squaring the unit, not the quantity. Whether or not there is a space does not matter.
Yes one could figure out what the article meant, but that does not make his point invalid.
Also, people have sided against you, because you are the only one defending the way it was written and people have dropped your score from 2 to 1.