Scott_Busse@mindlink.UUCP (Scott Busse) (03/07/91)
I put a faster oscillator (24 MHz) on my 2620, and changed the jumper to clock the math chip, but this is for the math chip only. This gave me a 10% speed increase on ray trace times, so for $7.00, it was worth it. NOTE that doing this puts your math chip at slight risk of overheating I suppose, so do so at your own risk. As for speeding up the '020, I don't think that is possible. 0 | \ ________Scott Busse /\_^_| |__ /_________CIS 73040,2114 `__ ----- \____/___________a763@mindlink.UUCP /\ _|_ \____________uunet!van-bc!rsoft!mindlink!Scott_Busse
aliu@aludra.usc.edu (Alex C. Liu) (03/07/91)
Hi, I have a neato 2620 board. I was wondering, it is possible to make it run faster. (Not just overclocking, but also replacing the chips with faster version, are there 28Mhz of 68020/68881/68851 chips?) I am aware that the 2620 is a synchronous board, so probably I will have to increase only in multiples of 7.44 (or is it 14Mhz?) but is it possible? Is so how? [Please dont send me suggestions about dropping the 2620 (9.8m/s^2 acceleration?) and getting a 68030 based board] BTW, is the 68882 pin compatible with the 68881? What are the differences? Thanks in advance -Alejandro Liu (PS: All mispellings are intentional!)
krauss@mansw1.enet.dec.com (Michael Krauss) (03/09/91)
In article <15562@chaph.usc.edu>, aliu@aludra.usc.edu (Alex C. Liu) writes: > >Hi, I have a neato 2620 board. I was wondering, it is possible to >make it run faster. (Not just overclocking, but also replacing the >chips with faster version, are there 28Mhz of 68020/68881/68851 >chips?) > >I am aware that the 2620 is a synchronous board, so probably I will >have to increase only in multiples of 7.44 (or is it 14Mhz?) but is it >possible? Is so how? > >[Please dont send me suggestions about dropping the 2620 (9.8m/s^2 >acceleration?) and getting a 68030 based board] > >BTW, is the 68882 pin compatible with the 68881? What are the >differences? yes, it is pin compatible. if your 68881 is in socket, it's jutst plug'n'play. the 68882 will give you somthing aroun 50% more speed with fp-calculations at the same clock rate. i think there is a jumper on the board to clock it async. to the 68020. if you can manage to use a new crystal you can buy a 50mhz 68882 and put it in. perhaps it'll work(depends on the rest of the board ?). has anybody used already such a configuration ? > >Thanks in advance > >-Alejandro Liu >(PS: All mispellings are intentional!) > -- ============================================================================== Michael Krauss #include <std_disclaimer.h> DEC Mannheim, Germany "AMIGA or VAX, nothing else please !" "krauss@mansw1.enet.dec.com" or "...!decwrl!mansw1.enet!krauss"
daveh@cbmvax.commodore.com (Dave Haynie) (03/09/91)
In article <15562@chaph.usc.edu> aliu@aludra.usc.edu (Alex C. Liu) writes: >Hi, I have a neato 2620 board. I was wondering, it is possible to >make it run faster. (Not just overclocking, but also replacing the >chips with faster version, are there 28Mhz of 68020/68881/68851 >chips?) Not very easily. You would absolutely have to keep 68020 and 68851 at the same clock speed, since they are very tightly coupled. The DRAM control logic would have to be changed, and the interface to the A2000 bus would really have to go asynchronous. >I am aware that the 2620 is a synchronous board, so probably I will >have to increase only in multiples of 7.44 (or is it 14Mhz?) but is it >possible? Is so how? See, being "synchronous" in Amiga coprocessor board parlance implies that the system is absolutely synchronized to the A2000 clocks. There is no easy you can go beyond 14.3MHz and still be synchronized. It might be possible with some kind of phase-locked loop clock gizmo, but that's not the only problem. Since it wasn't designed to handle asynchronous operation, the A2620 counts on certain setup and hold times, which may not be there even in a faster synchronous system. Many PAL speeds would have to be increaded as well (not a big deal if you're redesigning them anyway...). >[Please dont send me suggestions about dropping the 2620 (9.8m/s^2 >acceleration?) and getting a 68030 based board] Well, that's what I did. By the time the A2620 board was about done, I had a pretty good idea of how to make things run asynchronously, but I didn't try it out on the A2620 because of the complexity. It also helped that we were expecting to get 68030 samples at the time, so I took that concept, the A2620, the 68030, and produced an asynchronous A2630. >BTW, is the 68882 pin compatible with the 68881? What are the differences? The only differences are internal; all hardware signals are identical. Also, the 68882 is available in speeds up to 50MHz, whereas the 68881 basically maxes out at 16MHz (they used to have small quantities at 25MHz, but nothing in production volumes). I would imagine the 68881 is pretty much on the way out; Motorola can probably make the 68882 for as much, or even less, than the '881 by now. And the 68882 is a tad faster than the 80387 at the same clock speed, while the 68881 is a tad slower. >-Alejandro Liu -- Dave Haynie Commodore-Amiga (Amiga 3000) "The Crew That Never Rests" {uunet|pyramid|rutgers}!cbmvax!daveh PLINK: hazy BIX: hazy "What works for me might work for you" -Jimmy Buffett