CPU comparisons sought

As unbiased as it gets have a look at

and find your way through the information there. Certified scores from a benchmark consortium based on application oriented code.

Cheers, Schwob

Heres part of what I originally posted: "Is there any source out there that compares these CPU cores for efficiency, and relative strengths as well as say pipelines, bandwidth, floating performance and so on?"

This was because I knew everyone hates the word "best" so sarcasm is redundant.

I really wanted to see pointers for comparison of the cpu cores for their performance of various tasks, and their efficiency at it. For most FPU-based, bandwidth, raw integer, tight loops etc, x86 is not the most efficient (I'm not counting Athlon64 or AthlonXP here, this is for embedded).

So would you, REALLY consider the 4004 to be efficient, and for which tasks? Do you have references to back you up? Remember we're still not considering market share, availibility or software base here...

Sorry, never heard of it. The company I worked for made mobile phone systems (AMPS cellular was just starting out, there were still loads of non-cellular systems out there). The "PBX" wasn't a regular PBX: it sat between the base-station radios and the landlines. Most of the functions were the same as a PBX, and the loop-start and ground-start interfaces were pretty much what you'd find in a PBX, but the non-telco side interfaced to radio transceivers rather than telephone sets.

--
Grant Edwards                   grante             Yow!  World War Three can
                                  at               be averted by adherence
                               visi.com            to a strictly enforced
                                                   dress code!

code.

Don't believe the hype! Many of the benchmarks are as broken as Dhrystone.

Cheers, Jon

Not really there are a lot of cores 8/16/32 that do NOT have FPU, so the floating performance is to do with library efficiency and how the register set is organised and optimised for.

Lots of processors and controllers do not have pipelines. How good the bandwidth is depends on the application.

Without giving specific types of tasks the question is still meaningless as are so many benchmarks, which are more comparisons of compilers and target SYSTEM configuration. More often than not marketing gloss on product briefs and press releases.

One of its earliest demos and I believe application was traffic light controllers!

If we consider availabilty of software base there are LOTS of 8bit and 16 bit compilers, librarys and other bits of software out there.

Your question still seems to assume embedded = 32 bit .

Embedded is many things and somebody probably has some small form of analog computer or mechanical computing out there which is totally embedded. Consider the simple thermostat which can be considered a mechanical or even analog computing depending on the implementation. They are definitely embedded!!

To give an idea I use some 8 bit and 16 bit controllers for controlling and doing image processing on real time video, but they have

NO pipeline NO FPU MOST of the floating point is done by integer lookup tables Bandwidth is not an issue that is dealt with by hardware.

In answer to your question

"Twice the distance from the middle to the ends"

--
Paul Carpenter          | paul@pcserviceselectronics.co.uk
    PC Services
              GNU H8 & mailing list info
             For those web sites you hate

Yes, but it'd be easy to emulate with a modern FPGA. Is the microcode for the 4004 available license-free, or has anyone rev-enged it ?

Richard M Willis

I bet there is no microcode- just a bunch of gates.

Best regards, Spehro Pefhany

--
"it's the network..."                          "The Journey is the reward"
speff@interlog.com             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com

Probably true. On the outside chance that anyone cares, you can play with your very own 4004. The MCS-4 chipset (the 4004 was just the CPU component) has actually been implemented as part of an interesting simulation package in Java (see

) for the class documentation (in German) or for the main page for HADES. It's a very interesting project that might be appealing to some people here in comp.arch.embedded -- probably students in particular.

Ed

I'd be completely surprised if the 4004 had micro-code. I didn't think Intel did micro-code until the 386 (or maybe the

432?)

--
Grant Edwards                   grante             Yow!  Who wants some
                                  at               OYSTERS with SEN-SEN an'
                               visi.com            COOL WHIP?

I think it was much earlier. I remember hearing that the Z8000 (competition for the 8086) was delayed because it was "hard wired" in contrast to the 8086 organization.

--
Chuck F (cbfalconer@yahoo.com) (cbfalconer@worldnet.att.net)
   Available for consulting/temporary embedded and systems.
     USE worldnet address!

In NEC Corp. V. Intel Corp., the microcode of the V20 et al was claimed to be substantially similar to the microcode of the 8086/8088.

The concept of microcode predates the integrated circuit, apparently invented by this fellow:

Best regards, Spehro Pefhany

--
"it's the network..."                          "The Journey is the reward"
speff@interlog.com             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com

Maurice Wilkes. One of my most treasured books is one on computer design written by him and published in 1953!

Now there's an old geek.

Is there an english version of this anywhere ?

The 4004 is a good teaching example, and the hardware had many Low Pin Count features that are 'recently rediscovered'.

-jg

And the depth of the prefetch queue.