View Single Post
Old 07 January 2011, 22:45   #1476
Soldering Ironman
Oliver_A's Avatar
Join Date: Sep 2008
Location: Germany, NRW
Posts: 79
Originally Posted by gibs View Post

And I think that people in this forum want the maximum speed for the Amiga.
Yes, but not by driving the components completely out of spec. We spent most of the development time on getting the hardware completely stable. When it was possible to go up in speed (as done on the ACA630-30Mhz due to the faster SDRAM), we did so.

I will never hear someone saying : "Well... I'm happy with 25Mhz"...
I think that if you produce faster boards in the futur, people will be even more happy.
This is not the issue here. Again, if we had produced a 50Mhz ACA630, Cosmos would have overcranked it to 80Mhz.

Today you are doing a good job with the indy ECS but it's never enough...As you said for aDoom, the faster is the CPU the more you'll get FPS".
There is a difference between gaining speed through clever and efficient hardware design and gaining speed by pushing the components out of spec.

No he is not out of respect. definitely.
Well, if your answer to Jens detailed explanation WHY the SDRAM is clocked too fast is "All guys said exactly the same words...", then it shows some lack of respect, like as if Jens is inventing excuses and him being the clever guy who is exposing Jens. Just because his experiment works by accident and we don't even know how well it works. Again: memory corruption is very hard to detect.

We did develop the hardware. Not by an accident, but by carefully planning it out. We draw diagrams, we study datasheets, we take measurements, we optimize the logic code until the logic compiler reports that all timing specifications are met, we bust our behinds in hours of testing the design out with all kind of hard- and software combinations.

And if we say: 80Mhz is way out of spec, and even point to the specific parts of the memory design where this might fail badly (the Micron SDRAM datasheet is very easy to find on the net), I think we have a little bit more credibility than someone who is relying on proof by example ("but it works!").

I mean, everybody can do what he wants with his hardware. But we will respond to this here, because it's implying that we are cheating the people here by deliberately making hardware slower.
Oliver_A is offline  
Page generated in 0.07784 seconds with 10 queries