Why Do We Still Use Clock Frequencies? 45
Mr. Sketch asks: "With all the multiple pipelines, prefetching, caching, etc., that goes on in modern the (Bogo?)MIPS be a more accurate measure of a processors speed? If this is the case why don't chip manufacturers rate and advertise their chips with the MIPS value speed, but it seems like nowadays the MHz value is pretty much meaningless and we (as well as chip manufacturers) need to be using something else to get an accurate measure of the speed of a processor." I agree that clock frequency is next to meaningless when it comes to discussing the real speed of today's processors, but would MIPS really be a better replacement?
Re:I agree (Score:2)
Re:Marketing = Reality for many (Score:2)
Re:I agree (Score:2)
Arthur Dent thought that R17 was a bit too fast.
One horsepower equals (Score:2)
<O
( \
XPlay Tetris On Drugs [8m.com]!
Re:I agree (Score:1)
Re:Not a good way to measure CPU speed (Score:1)
Re:I agree (Score:1)
~Tim
--
For the love of god, don't use MIPS! (Score:4)
Measuring the performance of machines is way to complicated of an issue to use something like a MIPS rating. Because MIPS factors out the instruction count needed to get something done, you can inflate your rating by doing a large load of useless instructions really quickly.
MIPS = (InstructionCount) / (ExecutionTime *10^6)
ExecutionTime = (InstructionCount * AvgClockCyclesPerInstr * CycleTime)
The InstructionCount's cancel out, leaving
MIPS = 1 / (AvgClockCyclesPerInstr * CycleTime *10^6)
So if another computer can do the same amount of work with ten times less instructions, it doesn't show up in a MIPS rating.
Measuring performance just isn't as simple as looking at a single numeric rating. Sometimes you are interested in measuring responsiveness, somtimes throughput, and a lot of it depends on the specific applications you want to run. Just asking "how fast" is a processor is almost meaningless.
To my knowledge, the spec [specbench.org] benchmarks, while not free, are the best standardized benchmarks out there. For integer performance alone, it tests data compression (gzip and bzip I think), FPGA circuit placement, compiling c code, chess, running perl, ray tracing, database stuff, etc; I can't even remember all of the stuff it tests for floating point performance. Obviously, because it isn't free you probably won't be using it to test your home linux box, but if you are doing serious bench marks, the money would be worth it.
No single number is reliable (Score:1)
Anyone remember Cyrix? (Score:2)
This is always a problem... (Score:2)
I should say I am refering to rhe RC5 speeds...
http://n0cgi.distributed.net/speed/query.cgi?cp
Peace out.
Lots o' reasons (Score:2)
1) We're measuring instructions per second. No one's saying what the instruction IS. It could be a tiny one, could be a big one. Different instructions take different lengths of time.
2) You could get around this by getting an average MIPS based on the frequency of use of the various instructions, but that would mean you couldn't compare two computers with different instructions.
3) Some types of chips use longer instructions for efficiency. Now, technically, they are doing fewer instructions. But for all we know, those instructions could each be doing much more work than 1 "normal" instruction.
Truth is, neither MIPS nor Clock Cycle are that good a judge of speed. Measurement is just plain hard.
Re:Marketing != Reality (Score:1)
Re:For the love of god, don't use MIPS! (Score:2)
*sniff*....*sniff*
Hey, it smells like Hennessy and Patterson around here!
(We'll see if the mods get that one)You're correct, of course.
Re:Marketing != Reality (Score:1)
I mean they all have the same OS.
The only differences are RAM and processor.
But Apple managed to sell computer based on cuteness.
Re:Marketing != Reality (Score:1)
Re:Marketing != Reality (Score:2)
The problem is, we don't see each other cars this way. You don't open a car magazine and see a bunch ads reading
**240 HP!!!** $15999
***280 HP!!**** $18299
!*!*!*! 300 HP!! WOW !*!*!*! $23929
But that's exactly what computer magazines look like. If anything, it shows how unsophisticated the market for computers is, and that people are totally unaware of the 100s of small subjective things that actually sell things like cars.
The only real solution for this is for CPUs to get so fast that the differences are irrelevant (are we there yet?), much like how car horse power ceased being important in the 1930s (and again in the 1990s) and people can focus on the non-quantifiable bits like reliablity, stablity, and maintainablity.
Re:Agreement is necessary (Score:2)
Even then, only your own application is relevant. You might be doing multiple passes through an array -- but unfortunately you're sweeping memory addresses in a direction which is fast on one virtual memory system but requires a disk access for every calculation on another virtual memory system. (Yes, I've seen that happen.)
Lot of good comments here. (Score:2)
There have been a lot of good comments made by everyone, so I will only add a couple of my own.
As others have said, speed is best-calculated as whole system speed. Not only the speed of the components (hard drive RPM, for example) but the technology and implementation of that technology is a factor.
For example, is the hard drive 5400 or 10000?
Is it IDE? If so, how large is the cache? What is the transfer rate?
Even the fastest IDE will slow down a system with overhead, so lets talk SCSI.
Which SCSI? SCSI, SCSI-II, Fast SCSI-II, Wide or Ultra or Ultrawide or U2W or UW160 or Fiber Channel or....
Now, is it an elcheapo controller from CompUSA or is it a high-end busmastering controller? The difference is huge.
Okay, that's a lot of data about the disk system. For a server or a workstation it will make a huge difference. But if I am just cracking SETI packets it will make almost no difference.
See where I am going with this? There are too many variables to fit into any standard benchmark. The only decent way to test any computer is with your own software; the application you intend to be running.
Failing that, the best way for me--sadly--is the MHz of the chip. Why? If it is a decent (Intel, High-end Asus, SuperMicro) MB, the speed will be about the same from system-to-system. Cheap computers all use IDE disks that I will just turn into .MP3 archives,
and good computers usually all use the latest-and-greatest SCSI, often nice Adaptec
controllers. The MHz of the chip
gives me something to compare them with; two
similar configuration machines will be about the same speed cycle-to-cycle. As long as it is the same brand and model CPU! This does
not hold true for comparing a Laptop Pentium with a Pentium Pro, and comparing a Mac to an IBM is like comparing Apples to....you get the idea.
For real applications (servers) or across different platforms (HP700 vs UltraSparc vs Origin vs Alpha), most buyers know what they are getting into and run decent tests on the boxen.
Finally, for home users you can forget it; you can talk till you're blue in the face about why a Compaq DeskPro or Proliant will be faster than the $600 Compaq of the same MHz down at the local superstore, and they'll end up buying the $399 MediaGX E-machine anyway. Most home users only care about price. Change the 600Mhz sticker to 1200Mhz and they will swear it is faster...
I've learned my lesson on that one.....
Re:Sure, more MFLOPS wars... (Score:1)
Yes, and the first 3D video cards for home/office use were optimized for the (limited) kind of 3D stuff found in screen savers, rather than implementing all-round 3D acceleration. So Joe Blow could see that his fance new card made his cool 3D screen saver run really quickly, but it didn't say anything about what performance would be like with rendering or games or whatever.
Re:Marketing != Reality (Score:2)
iMacs don't sell 'cause they're 'cute'. Yeah, good design probably does enter into the equation (if you've gotta plunk down a couple grand for something that's gonna be in your living space would you prefer it look like a Singer sewing machine in it's case circa 1965 or something a bit more attractive?) but bang-for-the-buck they're decent machines.
Whoa now, DON'T go spitting up MHz or bus speeds or any of that - that only matters to hardware jockeys & game freaks. To the average customer the iMac is an accessible, reasonably powerful computer running the popular Mac OS at a reasonably affordable price. That's what it offers and that's why they've been selling like hotcakes.
It's not some damn Barbie clone walking into the store & squealing out "It's cuuute! I want one!" Nor is it Serge the designer saying "You must buy ze iMac - it goes with the Feng Shu of your home." At ~US$1500 a pop neither's a big market & it would have been exhausted long ago.
You want an iMac customer? Take my Dad: Prof. Mike Maggard, if you ever study business odds are you'll use his textbook in OM, he's one of the guys who pioneered computers in business. Dad's got an iMac on his desk at home. Why? 'Cause it does what he wants. Reads his email, browse the net, look at the occasional Word or Excel file, low maintenance, always on, and luggable enough he can drag it to the summer house with no hassle.
Problems with his iMac? None. It sits there quietly on it's table waiting to be used, when he moves the mouse it wakes up & is ready for action. Training to use it? None. At the university he has a WinNT PC and only wishes it was as reliable as his Mac.
Installed MS Office 2001 the other day - what was involved? Insert CD with Office 2001. Drag old Office 98 folder to Trash. Drag new Office 2001 folder from CD to HD. Double-click on any app, answer a few questions, import email, all done. Repeat: ALL DONE. No installer app crap, no hassle, 1 minute (longest part was copying from the CD) and he's cleanly upgraded his entire primary applications suite. Do that in NT? Do that in Linux? Thought not. Folks who want their computing experience to be this frictionless like Mac & the tutti-frutti colors are a bonus.
A very good analogy (Score:3)
So...for now we talk about MHz, MIPs, MBs, GBs, bits in a bus, bandwidth, frame rates and so forth because they are easy to measure, easy to quantify, and we really don't know what it takes to make a good computer.
In some places, mostly the embedded world, you hardly ever hear of MHz. Who cares how many MHz the proc inside your camera is? Unless of course you want to build a Beowolf cluster of cameras
Good judgement comes from experience, and experience comes from bad judgement.
Multiple definition of MIPs (Score:2)
Marketing != Reality (Score:4)
Answer: Use a number that does in some vague way represent a speed difference and sounds really sexy, MHz. The more MHz you got the faster you are baby!
Sure your neighbor has a 700 MHz box but for only a couple grand you can buy this new 1.2 *GHz* box and wipe him off of the map! We sold you cars this way, why not computers? Reality - pshaw - who cares? Joe Sixpack knows thay want "MHz" and MHz we'll sell him.
AMD tried convincing folks their 300 MHZ was just as fast as Intel's 400 MHz chips (or whatever the exact speeds were.) Didn't work, "Processor Class" went away & when AMD surged in speeds they never looked back. Apple tries to convince everyone that their 500 MHz PowerPC is comparable to a 700 MHz Intel PIII and while it may well be no one cares - columnist after columnist sneers at Apple for it's poky 500 MHz (or dual 500 MHz) chips.
Big-iron folks know, mini-folks know, workstation folks know, but the general computer buyer doesn't know that there's a dozen or so variables that affect the speed of a consumer box & CPU speed is only one of them. Motherboard speed, RAM speed, cache size & speed, hard drive speed, so many basic issues affect the 'speed' of a computer but are ignored for the MHz rating.
So know you want to communicate this information to Joe Sixpack who just wants to come into the store, drop a few grand to get a fast box tricked out with today's must-have technology and be back home in an hour? Or to Savvy Shopper who's bought a dozen geek mags in the past week, read over every one yet still has no clue of what any term means & will want the 19 year-old community college part-time clerk to try & explain it all?
No slams here but it aint gonna happen. Folks know MHz, they understand MHz are faster, they want MHz (or now GHz.) Sure they might buy a 1.2 GHz machine with crappy slow RAM & a 5400 RPM hard drive but it's gonna be FAST 'cause it's *1.2* *GHz*!
"It goes to *11*, man! Not just "10" like everyone else but to *11*!"
The rest of us roll our eyes but hell, that's the way the world works. You're not going to find another intrinsic value that communicates the speed of a computer to the general public better then MHz and there's no chance of getting everyone to agree to an artificial one.
Re:Marketing != Reality (Score:2)
But the marketing department doesn't care if the Whizbang2000 is faster than the MegaFooFoo2000. Remember, technical merit has no bearing on marketing. The only thing that matters is that the consumer thinks one is better than the other.
That's why automobiles (and Windows) are identified by model year. A 2001 model must be better than a 2000 model, even though in reality they are 99% identical.
Marketing (Score:3)
But then Marketing has to have the ability to teach the buyer. Not the easiest task in the world for something so technical.
Clock speed is deathly easy to understand. Faster clock speed means faster computer. (Of course the clock speed argument breaks down when you look at different processors/different buses/supporting architectures/etc)
My main point is: clock speed is the easiest and most effective sell for marketing.
We geeks know the difference.
Re:Marketing (Score:2)
//rdj
Agreement is necessary (Score:4)
Consistency is key- if website A compares processors with suite X of speed tests, you can only compare that test with other things done with suite X (and probably only with website A). MHz is a much easier and quicker way to compare. Just one number. Just not an accurate number.
Just wait until they don't have a clock! ;- (Score:2)
It's coming. There is just no way to get "timing closure" (i.e. resolve all timing differences/issues) in modern CBL (clocked boolean logic) IC designs as fast as the clock is, at the features sizes out there and the number of gates they sport. Most asynchronous technologies solve a number of power and EMI/clock skew issues, but there is still in the ease of design and design reuse (most asynchronous technologies are more difficult than CBL).
Enter Theseus Logic [theseus.com]'s Null Convention Logic (NCL). A dual-rail logic implementation that has all the benefits of traditional async, but also sports an inheritly delay insensitive nature and complete reuseable design at new feature sizes, temperatures and voltages. And unlike other asyncs, any CBL designer can be easily retrained to understand NCL. IMHO, NCL is the only viable solution right now that will solve the upcoming brick wall that will hit the CBL world by 2006.
Now since I've talked about NCL in at least 5 other /. posts, I'll let you read more. I'm no NCL expert (just the sysadmin at Theseus that seconds as a support engineer), so hit the web site for the most detailed info.
-- Bryan "TheBS" Smith
Not a good way to measure CPU speed (Score:4)
Hey, these are consumers we're talking about here. (Score:2)
Were wishes fishes in the sea, we'd all be wishy-washy.
Re:No single number is reliable (Score:2)
Sure, more MFLOPS wars... (Score:2)
Compilers optimazers used to be rated (mayb still are, but now a says gcc is all I look at) by running bemchmarks. However it turned out that some venders wrote code to optimize best for the situations of the benchmarks, even though it ment a 1 in a million case was the best optimized while more useful optimizations are ignored (due to lack to time to impliment)
I seem to recall a video card maker that detected benchmarks being run on the card, and didn't do anything other then report it was did what was asked.
I agree (Score:3)
Re:For the love of god, don't use MIPS! (Score:1)
Re:Marketing != Reality (Score:2)
I was talking to the owner of the local computer store the other day, and he can't sell anything under 5-600MHz in quantities that make them worth stocking, even to ppl who just want to surf the net, and do a bit of word-processing...
Will Joe Sixpack ever notice the difference between his neighbor's 800MHz and his 1.1G?
Doubtful.
Should they stop making them faster?
Anyone who's ever waited 24-hr to find out they misplaced a decimal point doing some number crunching (I hope I'm not the only one) will gladly say:
Bring it on! - I don't care if they sell them by MHz, MIPS, or by the Oz, just as long as it's faster...
Re:Marketing (Score:2)
My main point is: clock speed is the easiest and most effective sell for marketing.
I agree 100%. As much as they may irk and irritate us, the marketing department isn't stupid. They know enough about the technology to sell it, but more importantly, they know enough about the lowest common denominator, the consumer, to know what they're simply not smart enough to grasp.
Don't believe me? Click here. [techtales.com] I rest my case.
Bullshit! MHZ works fine in rating speed (Score:1)
Re:Marketing (Score:1)
Anyways, the point is that people buy cars based on horsepower because that's what marketing has taught us to do. Torque wins races, HP sells cars.
Just ask Mr. BigBlockMopar... doesn't matter if he's got a 426 or a 440, it'll still whoop a little honda with a big exhaust.
so why not be a bitch? (Score:2)
--
Peace,
Lord Omlette
ICQ# 77863057
Re:Marketing != Reality (Score:2)
Didn't hurt that Mercedes was the unquestionable top of the heap tho..
Re:Optimizer wars, and faking it (Score:2)
No cite, although I could probably find one if anyone cares enough.
Re:Marketing? Meet cuecat (Score:2)
> but i still to this day cant figure how you could think a peice
> of paper interacts with windows in that way......
Perhaps they just forgot to load the drivers for their barcode reader? (evil grin)
Marketing = Reality for many (Score:1)
Screw MIPS but... (Score:1)
When we reach some GHZ limit and moores law breaks (if ever) then maybe architecture and other factors will take over in deciding system speed but today the MHZ is the primary factor in differentiating a system five years ago and a system today.
Read more... [slashdot.org]