Ads watchdog underclocks reseller's 9.2GHz AMD CPU claim PC supplier Valatech - operator of website pconestopshop.com - has had its wrist slapped for claiming in an advertisement that a 2.3GHz quad-core processor runs at an overall speed fo 9.2GHz. The ad drew a single complaint which maintained that a running speed of 9.2GHz was not technically possible to achieve with a 2.3GHz …
There are always hundreds of listings for PCs on eBay that advertise clock speeds in this way. I reported it to eBay on a few occasions (pointing out that AMD don't actually make a CPU that runs at 9.2 GHz), but in typical eBay fashion, they did nothing.
It's goos to see someone finally getting a slap for it.
You see it all the time on ebay and occasionally in ads in the computer press.
My favourite overstatement (blatant lie) is speaker output, the best I have seen lately was a pair of USB powered speakers with a claimed output of 650 watts.
I once saw a set claiming 1000 watts output (2 inches high in several places on the box) which were powered by a 6VA wall wart.
I challenged the claim with the retailer (a high street name) and was told that speaker watts and electrical watts were different!
What price the laws of thermodynamics.
GHz ratings are pretty meaningless in isolation anyway, but if you wanted to add them up in that way the you'd want to chop off around 20% (I'm sure somebody can supply a better figure) when combining the individual core speeds.
eBay studiously ignore my many complaints about knock-off 'Google phones', 'G2s' etc, they're clearly too busy making money off the back of these criminal enterprises to bother policing them properly.
for the average user anyway. Personally I'd much prefer a 4GHz dual core to a 3GHz quad core. I don't want to do 4 things at a medium speed, I want to do one thing fast and not freeze the computer while I do it. Hence, all I need is a dual core.
Your average user probably doesn't even need that, at least not til office 15 when word will take up an entire 3GHz core to run the paperclip in full 3d.
<ot>
Actually, I think the guy in the shop was right. The "best" way to rate speakers is watts in RMS (root mean square) this is the figure you'll get from a proper audio company. The silly ones that retailler was quoting would have been MPO or PMPO (Music Power Output or Peak Music Power Output) which are fantasy football figures and the only way you'll get 1000 watts out of a device the size of a deck of cards.
</ot>
Paris, because she'd never mislead a potential pc customer.
...is that it requires a lot of expert knowledge. I mean, it seems logical to assume that 4 x 2.3 GHz cores are the equivalent of a single 9.2 GHz core. In truth, yeah its nothing like that, and is quite misleading.
It does show how much your average high street store is willing to mislead, sorry I meant lie, to get sales. They will tell you anything you want to hear to make the sale. Anything. I'm so glad I know enough about this sort of thing that I can make my own buying decisions and build machines myself, saves me a lot of time and money. Not necessarily effort, but I enjoy that part anyway.
If one ship can cross the ocean in nine days, how many days does it take three ships to cross the ocean? Three days? Not hardly, but you'd be surprised how many "students" checked three days as the correct answer.
Oh well, always count your change folks, math-o-matics is not a strong suit now days.
HD manufacturers aren't misleading anyone. They're actually using the correct term for data storage which was amended a few years ago. 'kilobyte' at al has been redefined, the old version is now called 'kibibyte' (kilobinary byte); OSes haven't been updated to reflect this.
Yes it's irritating and confusing, but not misleading.
"PMPO (Music Power Output or Peak Music Power Output) ... the only way you'll get 1000 watts out of a device the size of a deck of cards."
Even that's not true. The maximum voltage output of the amplifiers can't drive the resistive (not reactive) loads of the speakers to that level.
In the given example assuming 2x2R loads, 650W requires over 25V - USB gives only 5V. Realistically, PMPO is really 25W.
Perhaps manufacturers have specified what the speakers can tolerate for 1ms (before permanent damage), dismissing the fact that the amplifiers can’t drive them that hard.
GAH!
I can't stand the idiots that claim the HDD manufacturers are getting away with this trick, they're not, infact, they're about the only people that get it right.
the amount of people that don't know the difference between bits and bytes baffles me. It really does, but only because they think they know! get your facts right.
Surely all they had to do was advertise it as a "9.2GHz* PC" and in the (very) small print clarify that it wasn't actually a processor running at 9.2GHz but rather 4 individual cores each running at 2.3Ghz. Then, technically, it wouldn't have been misleading at all.
After all, the asterisk has never prevented phone companies from offering "Unlimited*" deals (which are clearly anything but) so why has this computer vendor been cloberred for a "misleading" advert when the phone companies get away with it every day?
<quote> If one ship can cross the ocean in nine days, how many days does it take three ships to cross the ocean? Three days? Not hardly, but you'd be surprised how many "students" checked three days as the correct answer. </quote>
Actually, 3 days *could* be the correct answer. If ship A starts at the start point, ship B starts 1/3 of the way there, ship C then 2/3rds. When you then start, 3 days later a ship will have crossed every point on the journey within 3 days.
"Although Valatech made reference to clock speeds, we considered that most consumers would understand that reference to 9.2GHz in the ad to be a statement about the overall operating speed of the computer," the ASA stated in its ruling.
So the ASA's got their motivation backwards.
Four cores, each running at a clock speed of 2.3GHz, would still be an overall system clock speed of 2.3GHz, because the clock is not dependent on the number of cores. Overall operating speed, however, is, so a 4-core CPU with 100% scheduling efficiency could have an overall operating speed of 4x a single-core CPU. So, the 4x multiplier is irrelevant from a clock speed perspective, but theoretically relevant from an overall system speed perspective.
The REAL problems with this ad are:
1. Operating speed != GHz, so they're using the wrong number to begin with.
2. Scheduling efficiency won't ever get to 100% in real-world applications, so the multiplication factor is wrong, and, of course
3. There are very few single-core solutions in the mainstream PC* market anymore, so there's no valid comparison anyway.
Right result, wrong reasons, eh?
* Netbooks don't count.
"the amount of people that don't know the difference between bits and bytes baffles me. It really does"
that sounds more to do with braodband download speed measurement than the hard drive 1000 / 1024 thing which has little to do with bits and bytes.
and btw , rms is about 0.707 of peak for a sinusoidal wave
Doug: Best analogy I could possibly imagine to explain the difference between clock speed and number of operations per second. I would have said it was fool-proof, but ...
Joe: No, that's just nonsense. What you have there is three ships, not a single one of which has crossed an ocean. You were asked how long for THREE ships to cross the ocean, not ONE ship anyway. The icon's for you, not Doug.
"Actually, 3 days *could* be the correct answer. If ship A starts at the start point, ship B starts 1/3 of the way there, ship C then 2/3rds. When you then start, 3 days later a ship will have crossed every point on the journey"
Assuming you are running a multi-threaded shipping application of course.
Q: How many M$ engineers does it take to change a light bulb?
A: None, they'd just redefine darkness as the industry standard.
HDD makers are right because they "redefined" the old "correct" way of listing storage with their very own NewSpeak version. Therefore they are right.
After all, I just redefined 4, so now 1+1=5, and I am right too.
"Actually, 3 days *could* be the correct answer. If ship A starts at the start point, ship B starts 1/3 of the way there, ship C then 2/3rds. When you then start, 3 days later a ship will have crossed every point on the journey within 3 days."
If the question was "how long does it take 3 ships to cover a third of the distance", then yes, but that's not really the question is it? Presumably because the Royal Mail delivers millions of letters per day, then each one takes a tenth of a second to deliver?
Though I'm charitably assuming you're being rogue-ishly tongue-in-cheek.
How long does it take to dig half a hole though?
"HDD makers are right because they "redefined" the old "correct" way"
The IEC redefined it, they're nothing to do with HDD manufacturers. This was done to harmonise the usage of the prefixes kilo, mega, giga etc.
Think about it, a kilogram isn't 1024 grammes is it; a kilovolt isn't 1024V is it!
"Actually, 3 days *could* be the correct answer. If ship A starts at the start point, ship B starts 1/3 of the way there, ship C then 2/3rds. When you then start, 3 days later a ship will have crossed every point on the journey within 3 days."
Ok how about this. All 3 ships dont leave at the same time. Each leaves at a designated time each day. Day 1 ship 1 leaves Day 2 Ship 1 arrives while Ship 2 leaves Day 3 Ship 2 arrives and ship 3 leaves......oh crap that means it takes 4 days to make 3 ships get across the ocean...........Maybe this is what the people writing the ad were thinking lol 4 cores/days 3days/less than half way to 3GHz :)
"rms is about 0.707 of peak for a sinusoidal wave"
Only for current OR voltage, not power - and that's with a resistive load.
"RMS is about 20% less than the maximum power."
No no no!
The 'root of the mean of the square' of a sinewave squared is half.
In english, the RMS power is 50% of the peak power (resistive load).
So if politics doesn't ultimately pan out for Sarah Palin, maybe she can start reeling in the big bucks by becoming an ad spokesmodel. She could not only deliver nonsensical statistics and figures without blinking to woo consumers, but could easily handle writing copy for them as well.
Palindrone - phrase that makes equal sense read backwards or forwards.
"If one ship can cross the ocean in nine days, how many days does it take three ships to cross the ocean? Three days?"
How big is this ocean and what direction are the ships going in and are they engine or sail powered? Well the wind could be blowing east to west and so a sail powered ship going west to east could take 9 days as it's fighting the wind and has to tack and so on. While these other 3 ships could be going east to west and so have the wind helping them and so make the crossing in 3 days.
<--- I'm assuming they are pirate ships.
You are all confusing hz with Speed when it should be thought of as Quantity, I know it sounds backwards but its the way it is.
A hz is a marble, the Cpu is a tube, it always takes 1 second for a marble to travel down the tube, its not how fast a marble can go down a tube, its how many marbles can travel down the tube in that 1 second that is the real speed of a CPU
lets take a Phenom II 965 X4 3.4 Ghz
This will give you access to 3.4 Ghz x 4 to give you 13.6 billion cpu cycles per second!
Thus giving you a theoretical 13.6 ghz processor.
A single core 13.6 Ghz CPU can handle 13.6 Billion Marbles per second
A Quad Core 3.4 Ghz CPU can handle 13.6 Billion Marbles per second
When somone says that a Quadcore is a 13.6 ghz Cpu what they are saying is that the CPU can process 13.6 billion cycles a second in total.
I noticed a couple of people mention the 'issue' with hard drives where 1GB = 1000MB.
This is actually correct, as they advertise their hard drive sizes in Gigabytes, of which 1 = 1000000000 bytes and not Gibibytes where 1 = 1073741824 bytes.
You are actually getting a bigger 'gigabyte' for your money when you buy RAM, the hard drive manufacturers are using the term correctly and are not ripping anyone off.
Just to clear up any confusion. :-)
If you start a lot of ships with a frequency of one ship per day, then, after the first ship arrives (assuming they carry identical load...) it will seem to the observer who only sees the endpoints of the journey as if a ship makes the trip in one day. If you now start another ship (or two or such...)slightly offset in time, then it will look as if they take less and less time. ? What an idea! Let's call it 'superscalar pipelining'.
RMS, PMPO, jpabncilg (stands for: just picked a big number 'cause it looks good) is all a bit meaningless if you don't specify what is delivered at that power level. There are amplifiers out there that sport a kW per channel in a relatively small box, and they deliver it by PWMing something like 200kHz from a SPS with the audio signal. Loud, yes, nice, not. THD should be given with the power rating, as well as a curve comparing P to k and f. And then listen to it. After all, you don't blisten to the label, do you?
Oh, btw, some makers actually measure amps like this:
1. Volume to max
2. attach voltmeter to unloaded output.
3. 0 db white noise to the input.
4. measure fast and memorize.
5. same with ampmeter, only look faster.
6. multiply open output voltage with shorted output current = (lots of digits of your choice)
7. Profit (oops, wrong channel, sorry)
For many years users of Intel architecture have used "MHz" as a capacity metric. It as been common practice to speak of the capacity of an Intel processor as "x MHz". In these assertions x is arrived at by multiplying the number of cores by the clock speed in MHz. The metric was call MHz rather than the more correct units "Processor * MHz" . It appears that our advertiser simply continued this practice but scaled the clock speed to GHz rather than MHz.. My guess is that they made the unfortunate but common error of talking about speed rather than capacity in the text that used the metric.
This metric used nearly always overstates the capacity of a multicore processor compared to a single core processor, but nevertheless has been in use for some time. Also 2 to 4 cores on a single chip the metric is much closer the than for larger core counts or 2-4 core machines implemented with single core chips. This means that the metric is more valid for single chip comparisons than it was for the multichip comparisons for whcih it was once in common use.
While the semantics may indeed be a violation , the use of the metric has served the Intel community for many years and can be quite accurate for loads that work well on Intel system. Since there is no such thing as "universal capacity metric" for comparison among server platforms, the metric is probably as good, if optimistic, as any other "benchmark" used to represent capacity. The problem is that if we hold advertisers to this standard, we give too much credance to the ability of "certified standard benchmarks" to represent real work.
One other point of interest. If you look at "standard metrics", I think you will find that AMD chips have traditionally gotten more work done per GHz than the Intel made cousins, so comparison to Intel chips based on this metric can actually undersell rather than oversell the chip. SInce the rules are set up to put a check on hype, it seems (from a technical view) to be a rather silly and frivolous ruling.
""HDD makers are right because they "redefined" the old "correct" way"
The IEC redefined it, they're nothing to do with HDD manufacturers. This was done to harmonise the usage of the prefixes kilo, mega, giga etc.
Think about it, a kilogram isn't 1024 grammes is it; a kilovolt isn't 1024V is it!"
You make a good point, and I concede on kilograms and kilovolts, etc. However, the accepted wisdom in computing terms was that a kilobyte was 10^2 bytes. It was that way for 30 odd years before the IEC got involved in 1998. HDD manufacturers had been selling HDD's with capacities where kilo was base ten long before this. They redefined what a "computing kilobyte" was to artificially inflate in marketing terms the capacities of their disks. Every other component and most software, even today, still uses 10^2 as the base calculation because, among other things, it better represents the way a binary computer accesses things. Adding in kibi, mibi and gibi just made things more confusing, not less as the IEC intended. It would have been easier if HDD manufacturers had simply fallen in line with the rest of the computing world.
I mean, a byte is always a binary value, never base ten, so "Giga Binary Bytes" is a bit redundant, don't you think?
But that's just how I feel about. And don't forget, whilst a gigabyte is a billion bytes, is that an American or an English billion?!?!?!
(That was rhetorical and a joke!)
I've seen this stuff before. I don't remember the reseller, but back in the old days, when dual core chips first started getting popular, there was a british reseller ad I got via spam (had to look, just because I thought maybe aliens had landed and changed everything) that claimed a clock speed of 5.2GHz. Reading the fine print, I saw they were adding the 2 cores of a dual core 2.6MHz cpu. We laughed about it in our irc channel for awhile. And now it's making big news...maybe I should have written an article about it back then...
Quack...
"HDD manufacturers had been selling HDD's with capacities where kilo was base ten long before this."
Any manufacturers doing that were naughty, but I've pesonally never come across any HDs using that scheme before 1998 (then again, I didn't use them much before then anyway - lol).
I agree base 2 lends itself to easier low-level programming because of the nature of binary and hardware arrays, but that's what you get when wisdom is substituted by bureaucrats .
As it stands *today*, HD manufacturers aren't being deceptive with their capacities.
Steve 6,
RMS of any (pure) sine wave is (roughly) .707 of peak amplitude, be it voltage, current, power, velocity of chipmunk excretia, what have you.
If you are trying to get RMS power and have measured peak voltage (or current) and want RMS power (and know the load), *then* the v-squared (or I squared) will get you the 0.5 factor.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
