Re: [Clax86list] When Knuth and I were young - Part 1
- From: Tim Roberts <spamtrap@xxxxxxxxxx>
- Date: Tue, 22 May 2007 03:49:50 GMT
Charles Crayne <spamtrap@xxxxxxxxxx> wrote:
?
On Sun, 20 May 2007 20:16:54 -0500
dave <spamtrap@xxxxxxxxxx> wrote:
Now I suspect that my AMD Athlon64 desktop would blow away the old
Cray 6600 which I worked on in 1969-70..
Quite possibly so, on cpu power alone, but the microcomputer
architectures are still several decades behind on I/O bandwidth.
That's just not the case.
The CPU in the CDC 6600 had a 10 MHz clock. Since the Athlon64 is now
moving as many bits in a cycle (64) as the 6600 did (60), it's quite
accurate to say that a 3 GHz Athlon64 CPU is 300 times more powerful than
the 6600, which cost $3 million and typically timeshared 50 users.
I/O was handled by 10 separate "peripheral processing units", or PPUs. The
design of the PPU was practically identical to Intel's hyperthreading
concept; the 10 PPUs had 10 memories but only one processor. Thus, each
PPU actually ran at 1 MHz. The CDC 844 disk, commonly used on those
machines, had a transfer rate of just under 1 MB/s. Even the 4-headed
Hydra 895, CDC's most advanced mainframe disk, only did 3 MB/s.
No, a single PCIExpress lane absolutely blows the door off of any mainframe
I/O from that era.
--
Tim Roberts, timr@xxxxxxxxx
Providenza & Boekelheide, Inc.
.
- References:
- When Knuth and I were young - Part 1
- From: Charles Crayne
- Re: When Knuth and I were young - Part 1
- From: Terence
- Re: When Knuth and I were young - Part 1
- From: dave
- Re: [Clax86list] When Knuth and I were young - Part 1
- From: Charles Crayne
- When Knuth and I were young - Part 1
- Prev by Date: Re: When Knuth and I were young - Part 1
- Next by Date: Re: generating a.out format file with gcc
- Previous by thread: Re: [Clax86list] When Knuth and I were young - Part 1
- Next by thread: Re: When Knuth and I were young - Part 1
- Index(es):
Relevant Pages
|
