Better than the x800?

Mutant_Fruit

The x800 is a .13 micron design, and it seems to me, the smaller the micron, the better the chip is...

So would the x300 be better than the x800, since its .11 micron?

Woden

me thinks not look at the effect a smaller design had on intel's prescott, smaller design allows higher frequencies me thinks, like the 9600pro which was 0.13micron iirc, while the 9800 was 0.15micron, so we saw higher frequencies on the 9600 but it still is not as good as the 9800 due to it having half the pipelines and stuff

other factors come in other then the the design size is what i'm getting at

secret_squirrel

Short answer No.

The X800 has 16 pixel pipelines to the X300's 4 and 6 vertex shaders to the X300's 2.

The X600 is somewhere between the 2.

The reason that a smaller size die is usually better is that companies usually redesign their cores when moving to smaller processes.

Reading the blurb on the ati site suggests that the .11 micron process is a shrunk version of their .13 process rather than a 'true' .11 micron process. Althought dont ask me for the definition of a true .11 micron process! There was a story (on the register) recently about sony producing their new cell cpu on a .9 micron process, until some analyst spotted that they were using a shrunken version of their .130 process.

secret_squirrel

A big driver of process shrinking is cost.... an .11 micron core is smaller than the same core at .13 microns. since the silicon wafers are a fixed size this means you get more cores per wafer and therefore a lower price per core. This either means cheaper retail prices or bigger profits

The fact that its a lowend product (prolly comparable to 1/4-1/3 of an X800's performance) suggests that the .11 process might not be suitable for a complex gpu like the x800. Either the x300 is a simpler design than the x800 or its an x800 with a good number of flaws (and therefore disabled/broken pipelines ). If the latter then expect .11 x800's at some point.

Azza

I'm fairly certain the X300 is a mobile graphics card?

jesus_thats_gre

They could be trying the smaller manufacturing size on lower scale cards in order to minimise the effect any quirks could have on their rep, i.e. they may not be willing to test it on their highest end product in case it fails dratically, thus granting nvidea the lead again...

The 9600 uses a different manufacturing process than the 9800 for example, from memory the x800 uses the same manufacturing process as the x800.. The core of the 9600 is different to that in the 9800 anyway.

Actually just checked it, and the 9600 uses the .13 micron process.

astrofool

ATI like to test a new process on not as complex chips generally. This way if something goes wrong, it can be fixed easier, AND it allows their latest and greatest to be made using a known method that works.

If the new process does fail, they can underclock an older process chip to fill in the niche, if its a success, than they can fine tune it as needed, and use it for their next generation of high end chips.

Dempsey

Originally posted by Dataisgod
smaller design allows higher frequencies me thinks

Bingo. The smaller the design reduces the capacitance effect on the individual switching speed of a transistor allowing higher switching (i.e. higher frequencies) but not a main factor of overall performance. Other ways to improve performance than raising the clock frequencies like as said piplines and more shaders, etc