VLIW pre-history
|
From: Nick Maclaren on 12 May 2007 06:30 In article <qhiray5pik.fsf(a)ruckus.brouhaha.com>, Eric Smith <eric(a)brouhaha.com> writes: |> Nick Maclaren wrote: |> > And lots of microcode from then on - for example, many of the |> > IBM System/370 range (e.g. the 165) used VLIW for microcode. |> |> Do you mean that in some sense other than that it was |> horizontal microcode? Well, it did use a VLIW format, too .... Regards, Nick Maclaren.
From: Anne & Lynn Wheeler on 12 May 2007 10:45 Eric Smith <eric(a)brouhaha.com> writes: > I was under the impression that the 370/165 and 370/168 microarchitecture > was fairly similar to the 360/65 microarchitecture. Perhaps I'm > mistaken. I haven't seen non-trivial portions of the microcode for any > 360 or 370 other than the 360/30. 155, 165, 158, 168 (3830 disk controller) ... lots of high-end stuff from the 70s ... used horizontal microcode. wide instruction format that directly controlled various hardware functions with no interlocks ... aka start of transfer from one unit to another ... the programmer counted instructions/machine cycles ... from the time they initiated the operation ... until the time that the results should be available. 115, 125, 135, 145, etc ... was vertical microcode ... something that looked like more traditional sequential programming. the low-end and mid-range vertical microcode tended to characterize the avg. number of (microcode) instructions executed per 370 instruction i.e. ten. the high-end horizontal microcode tended to characterize the avg. number of machine cyles per 370 instruction (since some amount of stuff went on in parallel) ... i remember one of the 165 engineers making some statement that one of the microcode optimizations going from 165 to 168 was getting avg. machine cycle per 370 instruction from from 2.1 to 1.6. then the transition from 168 to 3033 (started out using 168 wiring diagram mapped to faster chip) got it down to approx. 1.0. misc. past posts mentioning 165 to 168 avg. machine cycles per instruction http://www.garlic.com/~lynn/96.html#23 Old IBM's http://www.garlic.com/~lynn/2000d.html#61 "all-out" vs less aggressive designs (was: Re: 36 to 32 bit transition) http://www.garlic.com/~lynn/2000d.html#82 "all-out" vs less aggressive designs (was: Re: 36 to 32 bit transition) http://www.garlic.com/~lynn/2003f.html#56 ECPS:VM DISPx instructions http://www.garlic.com/~lynn/2003m.html#37 S/360 undocumented instructions? http://www.garlic.com/~lynn/2004j.html#56 RFCs that reference MD5 http://www.garlic.com/~lynn/2005e.html#59 System/360; Hardwired vs. Microcoded http://www.garlic.com/~lynn/2005p.html#29 Documentation for the New Instructions for the z9 Processor http://www.garlic.com/~lynn/2006k.html#54 Hey! Keep Your Hands Out Of My Abstraction Layer! misc. past posts mentioning slow down going from 3830 (disk controller) horizontal microcode to 3880 vertical microcode (one of the justification for transition to vertical microcode was that the burden placed on horizontal microcode programmer was so enormous). http://www.garlic.com/~lynn/2000b.html#38 How to learn assembler language for OS/390 ? http://www.garlic.com/~lynn/2000c.html#75 Does the word "mainframe" still have a meaning?> http://www.garlic.com/~lynn/2000e.html#54 VLIW at IBM Research http://www.garlic.com/~lynn/2001h.html#28 checking some myths. http://www.garlic.com/~lynn/2001l.html#63 MVS History (all parts) http://www.garlic.com/~lynn/2002b.html#2 Microcode? (& index searching) http://www.garlic.com/~lynn/2002o.html#3 PLX http://www.garlic.com/~lynn/2003b.html#7 Disk drives as commodities. Was Re: Yamhill http://www.garlic.com/~lynn/2003m.html#43 S/360 undocumented instructions? http://www.garlic.com/~lynn/2003n.html#45 hung/zombie users ... long boring, wandering story http://www.garlic.com/~lynn/2004n.html#15 360 longevity, was RISCs too close to hardware? http://www.garlic.com/~lynn/2005t.html#50 non ECC http://www.garlic.com/~lynn/2006c.html#6 IBM 610 workstation computer http://www.garlic.com/~lynn/2006c.html#46 Hercules 3.04 announcement http://www.garlic.com/~lynn/2006g.html#0 IBM 3380 and 3880 maintenance docs needed http://www.garlic.com/~lynn/2006q.html#50 Was FORTRAN buggy? http://www.garlic.com/~lynn/2006r.html#36 REAL memory column in SDSF http://www.garlic.com/~lynn/2006s.html#32 Why magnetic drums was/are worse than disks ? http://www.garlic.com/~lynn/2006v.html#17 Ranking of non-IBM mainframe builders? http://www.garlic.com/~lynn/2006v.html#31 MB to Cyl Conversion http://www.garlic.com/~lynn/2007e.html#38 FBA rant http://www.garlic.com/~lynn/2007h.html#6 21st Century ISA goals? one of the things that amdahl introduced (for high-end horizontal microcode machines) was something called "microcode" ... which was sort of a special 370-mode microcode ... where the engineer could implement more complex machine features using much simpler "vertical" (370) microcode ... as opposed to the much more complex task of horizontal microcode http://www.garlic.com/~lynn/2002p.html#44 Linux paging http://www.garlic.com/~lynn/2002p.html#48 Linux paging http://www.garlic.com/~lynn/2003.html#9 Mainframe System Programmer/Administrator market demand? http://www.garlic.com/~lynn/2003.html#56 Wild hardware idea http://www.garlic.com/~lynn/2005d.html#59 Misuse of word "microcode" http://www.garlic.com/~lynn/2005d.html#60 Misuse of word "microcode" http://www.garlic.com/~lynn/2005h.html#24 Description of a new old-fashioned programming language http://www.garlic.com/~lynn/2005p.html#14 Multicores http://www.garlic.com/~lynn/2005p.html#29 Documentation for the New Instructions for the z9 Processor http://www.garlic.com/~lynn/2005u.html#40 POWER6 on zSeries? http://www.garlic.com/~lynn/2005u.html#43 POWER6 on zSeries? http://www.garlic.com/~lynn/2005u.html#48 POWER6 on zSeries? http://www.garlic.com/~lynn/2006b.html#38 blast from the past ... macrocode http://www.garlic.com/~lynn/2006c.html#9 Mainframe Jobs Going Away http://www.garlic.com/~lynn/2006j.html#32 Code density and performance? http://www.garlic.com/~lynn/2006j.html#35 Code density and performance? http://www.garlic.com/~lynn/2006m.html#39 Using different storage key's http://www.garlic.com/~lynn/2006p.html#42 old hypervisor email http://www.garlic.com/~lynn/2006u.html#33 Assembler question http://www.garlic.com/~lynn/2006u.html#34 Assembler question http://www.garlic.com/~lynn/2006v.html#20 Ranking of non-IBM mainframe builders? http://www.garlic.com/~lynn/2007b.html#1 How many 36-bit Unix ports in the old days? http://www.garlic.com/~lynn/2007d.html#3 Has anyone ever used self-modifying microcode? Would it even be useful? http://www.garlic.com/~lynn/2007d.html#9 Has anyone ever used self-modifying microcode? Would it even be useful?
From: krw on 12 May 2007 11:12 In article <qhiray5pik.fsf(a)ruckus.brouhaha.com>, eric(a)brouhaha.com says... > Nick Maclaren wrote: > > And lots of microcode from then on - for example, many of the > > IBM System/370 range (e.g. the 165) used VLIW for microcode. > > Do you mean that in some sense other than that it was > horizontal microcode? > > I was under the impression that the 370/165 and 370/168 microarchitecture > was fairly similar to the 360/65 microarchitecture. Perhaps I'm > mistaken. I haven't seen non-trivial portions of the microcode for any > 360 or 370 other than the 360/30. I believe the 3165/8 descended from the /85. I haven't either. ;-) -- Keith
From: John L on 12 May 2007 13:09 >And lots of microcode from then on - for example, many of the >IBM System/370 range (e.g. the 165) used VLIW for microcode. VLIW isn't just wide microcode with multiple functional units. It's also the compiler techniques with trace scheduling and speculative execution that let it keep large numbers of units busy. The top end Multiflow machine had 28 functional units directly controlled by VLIW code. My impression is that mainframe microcode has far fewer. R's, John
From: Nick Maclaren on 12 May 2007 14:30
In article <f24sd5$doj$2(a)gal.iecc.com>, johnl(a)iecc.com (John L) writes: |> >And lots of microcode from then on - for example, many of the |> >IBM System/370 range (e.g. the 165) used VLIW for microcode. |> |> VLIW isn't just wide microcode with multiple functional units. It's |> also the compiler techniques with trace scheduling and speculative |> execution that let it keep large numbers of units busy. That sounds a bit religious! Remember that "RISC isn't just a reduced instruction set"? |> The top end Multiflow machine had 28 functional units directly |> controlled by VLIW code. My impression is that mainframe microcode |> has far fewer. Yes. But at least some were definitely VLIW. Regards, Nick Maclaren. |