The Gigahertz Battle: How Do Today's CPUs Stack Up? amd bias

[dragonsprayer]

once again amd bias - why? this seems fair?

well its not! This article is designed to make amd look good!

This article only makes amd chips look better then they really are. A 2.4ghz amd chip is near the top - the all mighty fx-60 is only 2.6 ghz and barely clock to 3ghz. While a C2D at 2.93ghz easly clocks over 3.8ghz or a P4 2.4c clocks to 4.1ghz THG did that test (4.1ghz) what 3 years ago?

amd chips barely can get over 3ghz. Well p4's clock easily at 4.1-4.25ghz. The article even says clock speed is not a true measure - all this proves is that THG loves to make amd look good if they can!

THG thinking: "lets compare amd's almost fastest chips to intels mid stuff - that make make amd look good"

4 years ago you could buy a a p4 to hit 4ghz even today no amd chip can hit 4ghz. we all know core 2 duo can be pushed to around 4ghz.

testing a p4 at 2.4ghz to amd at 2.4ghz to c2d is a poor test of true potential of each chip.

THG: "AMD has tried to educate people not to judge a processor based on its clock speed ever since the introduction of the Athlon XP"

how about apples to apples, this is not. This is a comparison of amd's near fastest to intels mid to low end! if you had overclocked the chips too it would have been a better and more informative article.

 

[gOJDO]

I agree! Their insane bias has no limits!
They have compared a midrange P4(Norhwood) against a much newer and the best Athlon64 the Clawhammer 4000+ s939. At least they should have compared a P4 Galatin instead Northwood, which is also older than the first K8. And they should have compared it against a s754 Athlon64.

They are comparing the 90nm AMD K8 CPUs, which are known to be faster than the latest 65nm.
:roll:

:lol:

 

[dragonsprayer]

I agree! Their insane bias has no limits!
They are comparing the 90nm AMD K8 CPUs, which are known to be faster than the latest 65nm.
:roll:

:lol:

crank that amd 2.4 to say what? 2.8 then crank the p4 2.4 to what? 4.25? wow i wonder if that might shift the graphics a little

sneaky slipping in the 90nm - i missed that!~

 

[RichPLS]

I thought it was illustrative of productiveness of the clock cycle being portrayed... not how they measure at stock speeds, nor how fast you might be able to clock them... [/shrugs]

 

[Grimmy]

Agreed.. That's what I got from the article. :?

 

[fender22]

I thought it was illustrative of productiveness of the clock cycle being portrayed... not how they measure at stock speeds, nor how fast you might be able to clock them... [/shrugs]

kinda what i thought..

 

[YO_KID37]

amd chips barely can get over 3ghz.

Your Completely an Idiot, Athlon64's have hit up to 4.1Ghz that's a good ~35% faster than what you've been Spraying

http://valid.x86-secret.com/show_oc?id=94127

 

[r0ck]

Stable?

 

[YO_KID37]

Who give's a Shytt.. I just posted it to SPAM on Sprayer.. Lol anything goes when it comes to DRAGON bashing.. Most time At least BaronMatrix at least makes some good points with some believable facts. But Dragon Sprayer took THG as a subject he would get 35% on his report card for lack of information and Most of the times making stupidity. Ask DaSickNinja in his Post up above

 

[MU_Engineer]

They compared chips running at 2.40 GHz, keeping system busses at stock for the line of chips they represent.

1. Pentium 4 Northwood 2.4C
2. Core 2 Duo Allendale E4300 at 2.40 GHz (266x9)
3. Core 2 Duo Conroe X6800 at 2.40 GHz (266x9)
4. Core 2 Quadro QX6700 at 2.40 GHz (266x9)
5. Athlon 64 4000+ Clawhammer
6. Athlon 64 X2 4600+ Toledo
7. Athlon 64 X2 4600+ Windsor

At first glance, this looks dumb as different/better chips could be used. The E4300 has a stock speed of 1.80 GHz (200x9) and there's no 2 MB L2 C2D that runs at 2.40 GHz. The X6800 and QX6700 are underclocked and real 2.40 GHz chips- the E6600 and Q6600- exist. The A64 4000+ is a Clawhammer and not the better 90 nm San Diego and the 2.40 GHz X2 4800+ with 2x1 MB L2 would do better than the 2x512KB L2 4600+.

But once you look at it, these choices seem to be a bit smarter. The performance per clock and not raw performance is being tested, so no 4.15 GHz P4 overclocks (if the Northwood didn't undergo SNDS getting there!) The E4300 was used as that's a way to test 2 MB L2 C2Ds versus 4 MB L2 versions since its 9x multiplier leads to a 2.40 GHz clock rate at the same 1066 MHz FSB as the 2.40 GHz 4 MB L2 chips. The Core 2 Extremes were used as their multipliers were unlocked and I'd bet that the THG writer didn't want to hunt down or did not have an E6600 and Q6600 sitting around but had the X6800 and QX6700. The X2 4600+ instead of the X2 4800+ was likely picked because there were VERY few X2 4800+ Windsors ever shipped but a ton of 4600+'s did. They could have used the 939 Opteron 180 and AM2 Opteron 1216 for a 2.4 GHz 2x1MB L2 chip, but it was probably easier to get the 4600+s than the Opterons or they had the 4600+s lying around.

What I would have liked to have seen included with the tests were a couple other 2.4 GHz chips:

1. A socket 754 Athlon 64 like the Clawhammer 3700+ or the Newcastle or Venice 3400+s.
2. 2.4 GHz P4 Northwood A and B.
3. Socket AM2 Orleans 3800+ and Lima 3800+.

That would have rounded out the test nicely as it would have gotten most every functionally-different revision of a 2.4 GHz chip ever made.

You seem to want to do a "maximum performance at maximum overclocking for each chip generation" comparison. That would be neat to see, but that's not the point of this article and deserves its own.

 

[hassa]

I think you kinda missed the point of the article. It was NOT to show the fastest CPU at stock speed OR to show how much they OC.
It was designed to show what happened when run at a common clock speed - i.e. if the speed stays the same, what difference does the architecture make???

A user posted on one of the forums the other day about CPU's not progressing much in the last year or so because they were not faster (i.e. clock speed is lower then some of the P4s). This article shows that architecture is important as it has a huge impact. All else being equal, the higher clocked CPU will be faster.

I did not see any bias in the article since there was no "CPU XXX" is better then "CPU YYY" (i.e. AMD or Intel is better) - the only bias that could be perceived is stating the the latest platforms from AMD and Intel are better then the previous platforms. And this isn't really bias - this is just common sense from the results.

But in the end, Intel CPU's are currently better since they OC more. But that was not the point of the article.

 

[MU_Engineer]

The guy I was replying to missed the point. You pretty much just restated my response.

 

[hassa]

I was targetting the response at DragonSprayer.
Totally agreeing with MU_Engineer.
(I should stop hitting the closest "Reply" button when I get to the end of the thread.... My bad!!!!)

 

[capnbfg]

It would help if you actually read the article before complaining about it. MU_Engineer perfectly summed up everything that you clearly missed, so I won't reiterate all of that. What I will do is pose a question to you:

If the article was such a biased piece of garbage, why did they allow the Intel quad core to win almost every test?




...the correct answer is that it isn't biased, because they were testing performance per clock, and if you consider the scope of the article, it was a good read.

 

[kiwik]

Seriously, you need to get laid.