Path: utzoo!utgpu!jarvis.csri.toronto.edu!cs.utexas.edu!romp!auschs!awdprime!woan.austin.ibm.com!ron
From: ron@woan.austin.ibm.com (Ronald S. Woan/2100000)
Newsgroups: comp.arch
Subject: Re: IBM RISC
Keywords: what integer parallelism?
Message-ID: <1666@awdprime.UUCP>
Date: 27 Feb 90 22:57:49 GMT
References: <8064@pt.cs.cmu.edu> <7454@pdn.paradyne.com> <1653@awdprime.UUCP>
Sender: news@awdprime.UUCP
Reply-To: @cs.utexas.edu:ibmchs!auschs!woan.austin.ibm.com!ron
Organization: IBM-Austin, AWD
Lines: 40


In article <1653@awdprime.UUCP>, ron@woan.austin.ibm.com (Ronald S.
Woan/2100000) writes:
|> In article <7454@pdn.paradyne.com>, alan@oz.nm.paradyne.com (Alan
|> Lovejoy) writes:
|> |> You forgot at least one possibility, which an IBM spokesman claims
|> |> just happens to be the real reason: the compilers have not yet been
|> |> updated to take any real advantage of the integer instruction
|> |> paralellism provided by the CPU.  
|> 
|> I must have missed something here... What integer unit parallelism?
|> From what I understand, the integer unit is a single pipeline (5?
|> stages), so it can never do better than one instruction/cycle. To get
|> five/cycle, you need two branches, float add, float mult, and integer
|> operation. Only an amazing compiler can schedule an application to use
|> this frequently.

Some people within IBM have asked me to clarify my statements (I
didn't realize there was any confusion). I did not mean to say that
compiler technology did not greatly enhance non-floating point
application speed. It can clearly do so by proper scheduling of
branches with respect to integer operations; however, with a single
integer operation pipeline, you will never see more than one integer
operation/cycle. Also, it would take an amazing compiler to schedule
five concurrent operations regularly for any imaginable typical
application to maintain the peak execution figure that some people
have been quoting.

Also, someone wanted me to mention that when I said a floating point
add and multiply in parallel, I did not imply that these were seperate
ops but rather a float multadd operation (instruction).

					Sorry folks,
					Ron

+-----All Views Expressed Are My Own And Are Not Necessarily Shared By------+
+------------------------------My Employer----------------------------------+
+ Ronald S. Woan  (IBM VNET)WOAN AT AUSTIN, (AUSTIN)ron@woan.austin.ibm.com +
+ outside of IBM       @cs.utexas.edu:ibmchs!auschs!woan.austin.ibm.com!ron +
+ last resort                                        woan@peyote.cactus.org +