Originally posted by: thilan29
Again not if those PCB layers were needed. If you have one card that NEEDS 8 layers and one that NEEDS 10 layers and the manufacturer decides to put 2 extra layers on both cards...how can you say the one with 12 layers is for sure higher quality than the one with 10 layers? Judging the quality of 2 different cards by the number of PCBs is near impossible as there are so many different factors that can decide the quality.
You can't be serious with this. Again, whether or not the additional layers are needed is irrelevant, the fact there are additional layers used is an objective measure of build quality, its really that simple. If you put two cards next to each other and one had more layers on the PCB then it absolutely would indicate higher build quality regardless of how it performed or whether it needed the extra layers or not. To give a relevant example of this, if I placed an 8800GTX next to a 4850 or GTS 250, I could absolutely come to the conclusion the 8800GTX had higher build quality with its more complex PCB, even if performance of the 4850 and GTS 250 were better or similar.
For example, the power consumption of the 8800GTX, 9800GTX, and GTX 260 are actually very similar (from xbitlabs reviews) so maybe due to whatever components and layout they used on the 8800GTX and 9800GTX they required the 12 layers whereas the 260 might have needed more because of the different memory bus in order to keep signals clean. Comparing the SAME card with the SAME components and layout you might be able to judge the quality on PCB layers but doing it for different cards is difficult IF you are looking for the real answer.
Actually comparing those parts lends creedence to the board partner's claims about Nvidia's reference designs being overkill, particularly the 9800GTX as I already mentioned earlier. The part was shown to be perfectly functional on a lower end PCB with the G92 GTS, was upgraded to the high-end reference board with the 9800GTX and 9800GTX+, and was again significantly downgraded again on the GTS 250. The chip was essentially the same throughout with a die shrink mixed in, but clearly was able to function at nearly identical speeds and performance on a much lower quality PCB that required less power overall.
A much better measure of quality would be the safety/tolerance margins for the components used, but as I said you wouldn't know any of that unless you were part of the engineering team that designed the cards. And it's pretty obvious you don't have any of that information. If you do, please enlighten us.
Once again, none of that is necessary as the proof is in the pudding. We can definitively determine superior physical build quality based on components and processes used and we can verify based on actual parts and performance.
The parts they switched over to 55nm and different components (eg. 260 55nm) have fewer PCB layers so again, maybe with the newer components and lower power consumption they don't need the extra layers.
Or maybe they didn't need such advanced PCBs to begin with and Nvidia just chose to err on the side of build quality for more reliable performing parts. Again, refer to the links I already provided indicating as much. The 55nm parts consume similar power, perform at a higher level, yet accomplish this despite "inferior" build quality and components, which to me clearly shows the reference boards were over-engineered to begin with.
There was a thread a little while back that showed the failure rates from some vendors in Europe and the GTX280 and the 3870x2 were leading the pack at a 10% RMA rate with the GTX 260 close behind. So where's the quality? Just so you and everyone else is clear, I'm not saying ATI has better build quality (I don't have the knowledge to make a statement like that) but you made a blanket statement that nV has better build quality, which is obviously not accurate and one you can't possibly prove (I mean really prove in a technical manner, not just making assumptions based on what you think constitutes higher quality).
Actually I believe that list had the GTX 280 and the 3870X2 at 10% and the rest of the field at ~5%, but if I had to guess more complex, higher performing parts have a higher failure rate or incompatibility rate. For example, the additional 8-pin connector over 2 x 6-pin by itself is a significant roadblock as it can often be the difference between a sub-500W PSU being sufficient or not. That additional 75W supplied and additional 12V rail amperage can often determine whether or not a PSU is sufficient.
Facebook
Twitter