home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.sys.amiga.misc,comp.sys.amiga.hardware,comp.sys.amiga.advocacy
- Path: wombat.hanse.de!root
- From: root@wombat.hanse.de (Bernd "Bernie" Meyer)
- Subject: Re: Motorola pricelist (was RE: CPU MHz in A1200+)
- Organization: Private uucp site lost in the voids of the black hole Germany is in the Internet
- Message-ID: <DL8FDx.140@wombat.hanse.de>
- References: <wfblanDKJ8Cw.1tn@netcom.com> <4car2m$76h@serpens.rhein.de> <DKtn25.MEC@news.zippo.com> <4cq2ui$j16@serpens.rhein.de> <4ctpl6$12h@dole.uninett.no> <4d0kep$ab6@murphy.servtech.com> <4d1dj9$cmd@serpens.rhein.de> <4d998s$1t8@murphy.servtech.com> <wfblanDL65sp.H0w@netcom.com>
- Date: Mon, 15 Jan 1996 17:03:32 GMT
-
- wfblan@netcom.com (Wells Fargo Bank) writes:
-
- >The bus which the PC's usually access memory from are only upwards of 33MHz
- >which 70ns RAM is plenty capable of handling.
-
- Make that "which 60 ns isn't any less incapable of handling". Accessing
- RAM with a 33 MHz cycle would require 30ns RAM (actually, it would
- require even faster RAM, as there is circuit delay, setup time etc.).
- On a 33 MHz bus, both 60ns and 70ns fast page RAM has a 4-2-2-2 burst
- access. With 60ns EDO, you will get 4-1-1-1, however.
-
- >Also, keep in mind that a 486DX2-66 is only being accessed from the Mother
- >Board at 33MHz...
-
- true.
-
- > Still within the 70ns range.
-
- See above.
-
- >Same with 486DX4-100, but its only 25MHz in this case.
-
- This, however, is utter nonsense. the "DX4" designation doesn't mean
- clock-quadrupling, although it would be logical. A DX4 CPU can either
- double or triple the external clock --- so a DX4-100 has either 33 or
- 50MHz external.
-
-
- >A 486DX4-120 is 30MHz and so on.
-
- Nope, that's the big problem for the DX4-120 --- it isn't able to work
- with a PCI bus (or at least it is rather expensive to make it work with
- one). Which is why AMD has brought out a new CPU (X5 or something like
- that) with a larger cache, better inner workings and clock-quadrupling,
- which will internally run at 133MHz and externally look like a 486 at
- 33MHz. It gives the P75 a run for the money....
-
- >The only possible exception I am
- >aware of is the AMD (I believe) 486DX2-80 which would access at 40MHz
- >and would seemingly be able to take advantage of 60ns RAM.
-
- Yep, the DX2-80, the DX50 and the DX4-100 at clockdoubling, and the
- DX4-120
-
- >Obviously, I am not aware of how this compares in the Pentium realm,
- >but I still only see systems selling with 70ns RAM with an occasional
- >60ns, again, is it taking advantage of the faster RAM?
-
- The Pentium 66/100/133/166 works with a 64 bit 66MHz bus. With 70ns
- fast page DRAM, you will get an 8-4-4-4 burst. With 60ns RAM, you can
- get 7-3-3-3, and with 60ns EDO RAM, 7-2-2-2. However, I am running my
- 60ns fast page DRAM at EDO rates, and it works without a flaw.
-
- >Basically, it appears that the Amiga
- >has the ability to access RAM faster
- >IF a faster CPU is being used
- >than the typical PC does.
-
- Well, a 40MHz and especially a 50MHz external memory interface cries out
- for 60ns RAM. However, the 40MHz version will normally work with 70ns,
- although it's not on the safe side. The 50MHz version will need another
- wait state with either, but the chances get away without are better with
- 60ns RAM, of course.
-
- However, I wouldn't say the Amiga's memory acces gets "faster" when
- you upgrade the CPU, it gets "less slower"
-
- Compile and time the following code on any Amiga with cache to test
- memory performance. My system (a PC) takes 5.35 seconds, for a memory
- bandwidth of 75M/s.
-
- Bernie
-
- ====
- #include <stdio.h>
-
- #define COUNT 1000000
-
- main()
- {
- int a=0;
- unsigned int i,j;
- int b[COUNT];
-
- printf("%d\n",sizeof(a));
-
- for (i=0;i<COUNT;i++) {
- b[i]=i;
- }
-
- for (j=0;j<100;j++) {
- for (i=0;i<COUNT;i++) {
- a+=b[i];
- }
- }
- printf("%d: %d\n",j,a);
- }
- ====
-
- --
- ==============================================================================
- I'm using an experimental version of "nn" to read news. So if anything seems
- strange about the headers of this mail or post, please inform me. Thanks!
-