Chris, > I guess you missed the point. Well, let's work through it shall we. > The programming time is > limited by the internal operation of the LPC CPU. Sure. I understand that. > Erasing sectors is time controlled by the CPU. The jtag link has > nothing to do with this. Those time delays are the same for > any programmer. Just try doing some IAP yourself from your > own internal code and you will understand real quick. You > can't do it any faster even from inside the CPU. I'm sorry if I'm a little thick here. You'll need to help me with the mathematics. Here's a little dump of what CrossWorks does when downloading using a CrossConnect: Downloading Loader to USB CrossConnect for ARM Programming 3.6 KB at 0x40000000 Programming completed in 141 ms -- 26,808 bytes/sec Downloading keil_mcb2140.elf to USB CrossConnect for ARM Erasing completed in 2.1 s -- 23,499 bytes/sec Programming 47.4 KB at 0x0 Programming completed in 1.5 s -- 32,575 bytes/sec Verifying keil_mcb2140.elf on USB CrossConnect for ARM Verifying completed in 375 ms -- 130,216 bytes/sec I downloaded 3.6K to RAM in 141ms, then I erased the LPC2k's flash in 2.1s, then I downloaded 47.5K to flash in 1.5s, and then I verified that it had downloaded correctly in 375ms. Total time from cold to debugging = 4.1s. With a wiggler: Downloading Loader to Macraigor Wiggler (20 Pin) Programming 3.6 KB at 0x40000000 Programming completed in 312 ms --12,320 bytes/sec Downloading keil_mcb2140.elf to Macraigor Wiggler (20 Pin) Erasing completed in 2.0 s -- 24,078 bytes/sec Programming 47.4 KB at 0x0 Programming completed in 2.9 s -- 16,859 bytes/sec Verifying keil_mcb2140.elf on Macraigor Wiggler (20 Pin) Verifying completed in 390 ms --126,317 bytes/sec So, I downloaded 3.6K to RAM in 312ms, then I erased the LPC2k's flash in 2.0s, then I downloaded 47.5K to flash in 2.9s, and then I verified that it had downloaded correctly in 390ms. Total time from cold to debugging = 5.6s. You said: "The download to the Flash/RAM only takes a couple seconds or less. But the programming of the flash takes about 6 seconds. That is due to the internal speed..." Yes, so eight seconds for 48K to download and program. That compares to 5.6s to download and program on the wiggler and 4.1s on the CrossConnect. So, where am I making my mistake, and why is the JTAGjet so much better? -- Paul.
Message
RE: [lpc2000] Slow OCD Remote/Insight debugging
2005-09-28 by Paul Curtis
Attachments
- No local attachments were found for this message.