DSPRelated.com
Forums

Doubt reg MIPS Vs MHz

Started by Unknown January 6, 2006
Hi

If a processor operating at 200MHz. Whats the Max. MIPS the processor
supports.
Is there a way to calculate MIPS from MHz ??
How do u relate MIPS and MHz ??



Regards
Shangai Sun



Ravikumar.R wrote:

> Hi > > If a processor operating at 200MHz. Whats the Max. MIPS the processor > supports. > Is there a way to calculate MIPS from MHz ?? > How do u relate MIPS and MHz ?? >
I have another question: why the number of idiots in this newsgroop had such a dramatic increase since n.y. 2006 ? VLV
Ravikumar.R wrote:

>Hi > >If a processor operating at 200MHz. Whats the Max. MIPS the processor >supports. >Is there a way to calculate MIPS from MHz ?? >How do u relate MIPS and MHz ?? > >
There is very little relationship between MHz and MIPs. Some machines execute several instructions in one clock cycle. Some take several clock cycles to perform one instruction. Even for a single doubling the clock speed may not double the frequency, due to memory speed constraints. The obvious related topic is "just what the heck is a MIP?". What most DSP people think of as an instruction tends to be something doing several times the work of what general software developers think of as an instruction. Regards, Steve
Ravikumar.R wrote:
> Hi > > If a processor operating at 200MHz. Whats the Max. MIPS the processor > supports.
If you processor executes one instruction per clock cycle then your 200MHz processor have 200MIPS (Million Instruction Per Second). gold
>Ravikumar.R wrote: > >> Hi >> >> If a processor operating at 200MHz. Whats the Max. MIPS the processor >> supports. >> Is there a way to calculate MIPS from MHz ?? >> How do u relate MIPS and MHz ?? >> > >I have another question: why the number of idiots in this newsgroop had >such a dramatic increase since n.y. 2006 ? > >VLV
What a dumb question! --Bhooshan
Please be kind to the newbies!  If you consider their questions to be "beneath 
you", simply ignore them rather than insulting their intelligence.  Perhaps 
someone else will take the time to respond and either answer the question or at 
least better direct them on asking the "right" question.  comp.dsp has a history 
of being polite and friendly, and it would be nice to keep it that way.  Thanks!

-- 
Jon Harris
SPAM blocker in place:
Remove 99 (but leave 7) to reply

"Vladimir Vassilevsky" <antispam_bogus@hotmail.com> wrote in message 
news:w4qvf.50198$7h7.7831@newssvr21.news.prodigy.com...
> > > Ravikumar.R wrote: > >> Hi >> >> If a processor operating at 200MHz. Whats the Max. MIPS the processor >> supports. >> Is there a way to calculate MIPS from MHz ?? >> How do u relate MIPS and MHz ?? >> > > I have another question: why the number of idiots in this newsgroop had such a > dramatic increase since n.y. 2006 ? > > VLV
"Steve Underwood" <steveu@dis.org> wrote in message 
news:dplbb8$hjg$1@home.itg.ti.com...
> Ravikumar.R wrote: > >>Hi >> >>If a processor operating at 200MHz. Whats the Max. MIPS the processor >>supports. >>Is there a way to calculate MIPS from MHz ?? >>How do u relate MIPS and MHz ?? >> > There is very little relationship between MHz and MIPs. Some machines execute > several instructions in one clock cycle. Some take several clock cycles to > perform one instruction. Even for a single doubling the clock speed may not > double the frequency, due to memory speed constraints. > > The obvious related topic is "just what the heck is a MIP?". What most DSP > people think of as an instruction tends to be something doing several times > the work of what general software developers think of as an instruction.
In many (most?) processors that are considered DSPs, every (or almost every) instruction takes 1 clock cycle. So in this case MIPS = MHz. This is typically _not_ the case for CPUs used in PCs (Pentium, AMD, etc.) where it takes a variable number of clock cycles for each instruction to execute.
>Please be kind to the newbies! If you consider their questions to be
"beneath
>you", simply ignore them rather than insulting their intelligence.
Perhaps
>someone else will take the time to respond and either answer the question
or at
>least better direct them on asking the "right" question. comp.dsp has a
history
>of being polite and friendly, and it would be nice to keep it that way.
Thanks!
> >-- >Jon Harris >SPAM blocker in place: >Remove 99 (but leave 7) to reply
Jon, niceties need never be preached, it is self evident if anyone bothers to hang around comp.dsp for a bit. Iam not sure if this person does not know this already. --Bhooshan
"Jon Harris" <jon99_harris7@hotmail.com> wrote in
news:t9wvf.13$Xo5.11@trnddc02: 

> "Steve Underwood" <steveu@dis.org> wrote in message > news:dplbb8$hjg$1@home.itg.ti.com... >> Ravikumar.R wrote: >> >>>Hi >>> >>>If a processor operating at 200MHz. Whats the Max. MIPS the processor >>>supports. >>>Is there a way to calculate MIPS from MHz ?? >>>How do u relate MIPS and MHz ?? >>> >> There is very little relationship between MHz and MIPs. Some machines >> execute several instructions in one clock cycle. Some take several >> clock cycles to perform one instruction. Even for a single doubling >> the clock speed may not double the frequency, due to memory speed >> constraints. >> >> The obvious related topic is "just what the heck is a MIP?". What >> most DSP people think of as an instruction tends to be something >> doing several times the work of what general software developers >> think of as an instruction. > > In many (most?) processors that are considered DSPs, every (or almost > every) instruction takes 1 clock cycle. So in this case MIPS = MHz. > This is typically _not_ the case for CPUs used in PCs (Pentium, AMD, > etc.) where it takes a variable number of clock cycles for each > instruction to execute. > > >
MIPs/MHz can even be a bit more misleading. Many DSPs (and other processors) use a much lower clock in frequency and multiply this clock up via an internal PLL. Therefore the MHz we are comparing is the Core Clock, which as Jon points out is usually the instruction rate. Most newer DSPs support SIMD as well. This means that a single instruction can operate on Multiple Data (typically 2 sets). This effectively doubles the computation power of the DSP when executing code written to take advantage of this capability. The basic housekeeping functions rarely can take advantage of SIMD so the DSP is not going to be twice as fast as a SISD (Single Instruction, Single Data) processor but will be somewhere between 1x-2x faster. DSP manufacturers also like to talk about other metrics to show that DSPs do a lot of operations in one instruction cycle. This is why you see specs like MFLOPS/sec that are in the billion/sec range. -- Al Clark Danville Signal Processing, Inc. -------------------------------------------------------------------- Purveyors of Fine DSP Hardware and other Cool Stuff Available at http://www.danvillesignal.com
In comp.dsp,
Jon Harris <jon99_harris7@hotmail.com> wrote:
>"Steve Underwood" <steveu@dis.org> wrote in message >news:dplbb8$hjg$1@home.itg.ti.com... >>> >> There is very little relationship between MHz and MIPs. Some machines execute >> several instructions in one clock cycle. Some take several clock cycles to >> perform one instruction. Even for a single doubling the clock speed may not >> double the frequency, due to memory speed constraints. >> >> The obvious related topic is "just what the heck is a MIP?". What most DSP >> people think of as an instruction tends to be something doing several times >> the work of what general software developers think of as an instruction. > >In many (most?) processors that are considered DSPs, every (or almost every) >instruction takes 1 clock cycle. So in this case MIPS = MHz. This is typically >_not_ the case for CPUs used in PCs (Pentium, AMD, etc.) where it takes a >variable number of clock cycles for each instruction to execute. >
And that brings us back to: "just what the heck is a MIP?" :-) I normally use MCU's and the first time a used a DSP (AD2100 series), I was surprised that it could use the ALU, MAC and Barrel shifter in a single 'instruction' and on top of that provides hardware ring buffers and zero overhead looping. Oh, and the serial stuff also keeps running. Try that on a 'normal' MCU and you'll need a series of instructions. Using an ARM7, I was pleasantly surprised by the MAC instruction that speeded up FIR processing. Most MCU's I worked with before did not even have that. -- Stef (remove caps, dashes and .invalid from e-mail address to reply by mail) Love is the triumph of imagination over intelligence. -- H. L. Mencken