| ??? 12/19/08 18:11 Read: times |
#161056 - times Responding to: ???'s previous message |
Richard Erlacher said:
Andy Peters said: Those are all applications in which a substantial interruption in input data flow is tolerable. If I have to receive a modulated signal, demodulate and decode it the PC, operate on it, modulate the result, and transmit it back to the channel corresponding with that from which it came, all within a predefined window, those pauses are totally intolerable.So you don't consider high-end video editing and rending, and digital audio production to be "useful work"? Get over yourself, please. When I am doing a live multitrack audio recording, pauses are intolerable. Similarly the folks capturing live video to disk cannot tolerate pauses. It's difficult to drive over a cheap cable, but since there are cables capable of being driven at 10 Gbps, it would not be inconceivable to drive 32 of them. That wouldn't necessarily be at 10 G cycles per second, but certainly at 25-50 MHz. They were able to do that with SCSI! And SCSI is dead, because the cables got really expensive, and it was finicky about termination. FireWire is superior to SCSI in pretty much all aspects. I certainly didn't mean to suggest that there are no useful products that use USB. Yes, lots of useful products connect via USB. I've not seen any that have to have the computer's attention for more than a fleeting moment, however. I'm thinking that it may take two USB channels, one in and one out, and a really large FIFO buffer to compensate for those half-minute pauses. Funny, I do streaming audio using USB every day, and low-latency audio at that, meaning small buffer sizes, meaning more CPU overhead. I don't understand your problems. ...but I liked the way things worked in '88 ... which, BTW, was before Windows3.x. Have you read Steven King's "Dark Tower" series? There are tons of references to the time "before the world moved on." And the world really has. I don't want to go back to a DOS machine. I like having multiple windows open on the screen, where I can edit my VHDL in one and simulate in another and have the Xilinx tools running in the third. I DO NOT want to go back to doing FPGAs in ViewLogic for DOS. BTW, we tried scan-rate conversion between a truly high-end video display system and NTSC, using a MAC, and we could do the job using a 68012 with the aid of a DSP, but not on the MAC of the time, which had a 68030 (?) and a DSP on a NuBUS card. The MAC implementation relied on software in place of hardware so much that it only ran at applications an equivalent rate of about 2 MHz. When we ran it on the 68012 at full speed, ( the 68012 had a few more address lines that the 68010 lacked ) we were able to make the conversion in real time. It was a onesie, and it worked to the customer's satisfaction.
That experience, in combination with many other experiences with Apple products, persuaded me to ignore Apple products in the future. What other experiences? Besides, you compared a hardware implementation to a software implementation and you're surprised that hardware is faster? They always seem to lean toward play rather than useful work. I haven't had any cause to change my mind ... yet. Like I said, the video and audio folks are not playing. The graphic designers are not playing. Face it, there are creative jobs other than microcontroller programming. It's sad that there's such a big market for that sort of thing. Those who like that stuff get what they deserve, I guess. I hope they enjoy it. You're just jealous because Apple's products sell in the millions, and yours don't. I just wish it didn't impact the PC vendors and Microsoft so much. The PC was, for a decade and a half, quite a useful tool. "was" a useful tool? Without the Dell on the desk here, I would not be able to do my job. Again: FPGA tools. Simulation tools. C compilers. Schematics. PCB layout. All done on a PC we bought six months ago and perfectly useful. And ya know what? DOS OrCAD 386+ was a buggy piece of shit. Good riddance. -a |
