Path: utzoo!utgpu!watserv1!watmath!att!pacbell!pacbell.com!decwrl!sun-barr!olivea!orc!inews!iwarp.intel.com!psueea!parsely!agora!billsey From: billsey@agora.uucp (Bill Seymour) Newsgroups: comp.sys.amiga.hardware Subject: Re: Difference between RAMs used on Amiga and Mac. Message-ID: <1990Sep10.232157.534@agora.uucp> Date: 10 Sep 90 23:21:57 GMT References: <3937@quiche.cs.mcgill.ca> Distribution: na Organization: Open Communication Forum Lines: 58 In article <3937@quiche.cs.mcgill.ca> tinyguy@quiche.cs.mcgill.ca (Yeo-Hoon BAE) writes: : : Hi, I am one curious Amiga user who has a question about type of ram :chips that computers use... : I understand that there's 16bit and 32bit rams available, and to get :the full performance out of Accelerators(020/030 based), you NEED 32 bit :memory. So far so good, BUT they cost much more than the average 16bit :chips used on 2000. So, ok, I guess it's 32bit, and it has faster access :time, so that's reasonable. Then I talked to this guy who used to own :Mac SE, and now he owns Mac SE/30(which is 030 based). He also claims :that there is no such a thing as 16/32 bit. On Mac, they all use the :same type of chips(SIMM), the only difference if that faster machine :requires faster memory. So why doesn't Mac require more expensive 32 bit :memory and still able to perform reasonably? Is it the way these :machines are designed? I heard that 32bit processors are VERY slow with :only 16bit memory. :Any help would be appreciated. The ram chips that are used both in 16 bit applications and 32 bit applications are the same chips. It's just the way the board is designed that's different. On a 16 bit board, there are 16 data lines going to those ram chips. On a 32 bit board there are 32 data lines going to the ram chips. If you design in chips that are one bit wide (256Kx1, 1Mx1, 4Mx1) then you will need to add chips in 16 chip increments for a 16 bit wide board and in 32 chip increments for a 32 bit wide board. You use correspondingly less chips if you design in chips that are 4 bits wide (64Kx4, 256Kx4, 1Mx4). Not that either way, you end up using the same total number of chips on the memory board. (ie., 32 1M chips for 8M or ram.) On a typical 32 bit wide board, there are a few more components than on a 16 bit wide board. That's because you have to buffer each of those data lines, and there are twice as many data lines on the 32 bit board. You also tend to build smaller numbers of board when you design 32 bit wide boards, since there is a much smaller user base for systems that are using your 68020/68030 card. You also tend to use faster chips, just for performance. On a 16 bit wide board, you're typically designing for the Zorro II bus, which runs at 7MHz. You can get by with 150ns chips and run at zero wait state at 7MHz. If you're trying to run at 25MHz, you won't get zero wait states even with 60ns chips! Note that five wait states at 25MHz is still faster than no wait states at 7MHz... Between lower volumes and faster chips, the costs are a lot higher for 32 bit wide board than for 16 bit wide boards. Even though the memory itself is the same cost, or with a very small difference, anyway. : : :************************************************************************* :* Yeo-Hoon Bae * Amiga /// * :* tinyguy@calvin.cs.mcgill.ca * 2000 /// * :* tinyguy@quiche.cs.mcgill.ca * \\\/// * :* Amiga2000 + 3MB + 48MB HD + KX-P1124 + DiamondScan * \XX/ * :************************************************************************* -- -Bill Seymour ...tektronix!reed!percival!agora!billsey ============================================================================= Bejed, Inc. NES, Inc. Northwest Amiga Group At Home Sometimes (503) 281-8153 (503) 246-9311 (503) 656-7393 BBS (503) 640-0842