THG's The Mother of All CPU Charts 2005/2006

Funny how intel wins a lot of the synthetic benchmarks, but gets trounced in real world performance.
 
Should be even more interesting in 2007 with 65nm, quad core and DDR3!
 
How nice of them to benchmark cutting edge games such as Quake 3 Arena and Enemy Territory. :rolleyes: Isn't it almost pointless to run both of those anyway, since they use the same engine?

I didn't see a single game on there that supported SMP (no Q4, no Serious Sam 2, etc). For an article that claims to be a huge CPU review, and given the fact that both companies are pushing dual core, that seems like a pretty sad omission.

They also did all of their benchmarks using a 6800GT. That’s great and all, but how are game benchmarks that are GPU limited supposed to tell us any real useful information about the different CPUs?
 
Tom's always been somewhat biasing towards Intel. However, this time around no matter how Intel tries to massage numbers it's not going to work in their favor.

He gets alot of hardware previews for Intel equipment first.
 
That makes it all the more impressive that AMD came out on top for all but one tests despite the THG/Intel bias. :cool:
 
Stoopid Guy said:
I enjoy how he/they split the chart when it shows AMD trouncing Intel, but keep it together when it's relatively close. Dang Intel bias ! :rolleyes:
Click to expand...
I enjoy how you form an opinion off a set of data you don't even bother to interpret. The charts are used only to introduce the concept of Power vs. Performance and show the general trend over the years. The term used in the titles is 'Performance' but it's only meant as a loose substitute for clock speed. Since we all know that clock speed and performance are, at best, only analogous when comparing within a company, there must be one chart per company.
Put simply, the charts are not relatable. If they had been combined, Intel would appear to be favored (careful what you wish for), and people who want to argue Tom's bias would actually have a case for it.


freeloader1969 said:
AMD wins in every category, but according to Tom's they're not stable. What a *ucking idiot.
Click to expand...
He doesn't state that. What he points out about chipsets is only fact, and he uses examples from his own (THG as a whole, rather) experience as backup. Everybody on this forum knows they can build a stable AMD rig from any processor available, which is why all of us hate to see (myself included) people pulling the old AMD chipset card. It's a very weak argument, but it isn't a lie. AMD makes great chips but the companies that support those chips do make mistakes. That's why there are so many chipsets, and especially revisions of chipsets, available to us. However, it's always smart to research something before you buy, and even if you did buy at random you aren't likely to experience something that will effect you.

I generally only read tech review sites when linked to from these forums, so I have no allegiance to THG or any other site. I read most of this article and actually thought it to favor AMD more than anything. For someone who is being blamed for loving Intel, he sure makes a lot of remarks that blatantly give AMD the thumbs up.

When you bias yourself against someone for being biased, you're no better than Tom.
 
Tom's? Reliable? Don't make me laugh.

This is a guy who insisted that it was physically impossible to overclock a Cyrix processor.

Despite being emailed detailed technical information on exactly how to adjust the jumpers on an Intel i430VX reference board, including the P55C voltage, to clock a Cyrix P166 at 200MHz.
A box I ran as a joke for many many moons.

The entire winter salt supply for the mid-west as a region is not sufficient when reading Tom's. It's an open secret that they're basically owned by Intel, who funds pretty much all their testing. I've never once been able to reproduce their supposed results, especially when it comes to high end CAD cards - something they are uniquely unqualified to even look at, forget 'benchmark.' (Excuse me, ever heard of ICDs? Nope, didn't think so...)
 
RawsonDR said:
When you bias yourself against someone for being biased, you're no better than Tom.
Click to expand...
well said rawson. i personally don't care for tom's myself, but i'm not gonna put myself in a compromising situation by putting words in someone's mouth ;)


9 :cool:
 
If Intel owns THG, why is AMD winning every benchmark?

Anyway... woot @ the multi-threaded benchmarks, dual core is just plain great.
 
He doesn't state that. What he points out about chipsets is only fact, and he uses examples from his own (THG as a whole, rather) experience as backup.
Click to expand...

What Athlon 64 chipset isn't stable? I've never ran into any problems with any A64 chipset, whether it be VIA, Nvidia or even ALI. They all run at stock speeds with no issues.


RawsonDR....I'm just quoting what Tom's said. Also, do you know of ANY A64 chipsets that don't work as intended to? I know VIA has had a checkered past with Athon XP/Thunderbird chipsets, but I can't think of any problems with A64 ones. Short of not having an AGP/PCI lock (which is only needed for overclocking) I can't think of any issues.

Quoted from the article...

As a result, AMD has had a difficult time providing a stable and mature desktop platform.
Click to expand...
 
guff.JPG



Just a bit of phun.
 
i just decided to read it for kicks

For example, the Athlon 64 3500+ exists with and without SSE3 extensions as well as in 90 nm and 130 nm versions. In other words, the customer may buy the Clawhammer, Newcastle, Winchester or Venice version of the processor, but wouldn't be able to tell the difference from the retail package or the model number. This is problematic, as all of these versions differ significantly where performance and heat dissipation are concerned.
Click to expand...

somebody shoot me. since where is there a "significant" difference between any of the 4 cores the 3500+ came in when it comes to performance? :confused:
also.. at any given time, it's pretty hard to find more than one core, aside from when it came in newcastle and clawhammer varients, which were identical for all practical purposes. there was so overlap between transitions, but i'll bet the winchester, newcastle and clawhammer 3500+s are basically non-existant :D


edit!

Socket 754 only has a memory interface.
Click to expand...

HAHAHA! i'm dying here :p


edit2:
i feel bad, but i love picking stuff like this apart


The Winchester was followed by the Venice stepping, which added SSE3 support to the Athlon 64 and lowered the power consumption even further.
Click to expand...
WRONG! they're about the same
 
[RCKY] Thor said:
Funny how intel wins a lot of the synthetic benchmarks, but gets trounced in real world performance.
Click to expand...

I don't think this is an accurate characterization. Any good benchmark is not a real world test and doesn't accurately reflect real world performance.

While a single core AMD might show a better score for a particular app (for the sake of argument) than its equivalently priced Intel single core, if the Intel has HyperThreading, its likely going to give real users a much nicer experience.

Benchmarks are great for establishing relative performance levels and for establishing optimal equipment choices at given price points. But they can't be the only guide since there are some signficant factors they rarely, if ever, can take into account.
 
Any good benchmark is not a real world test and doesn't accurately reflect real world performance.
Click to expand...

So the benchmarks built into games are totally useless then? Benchmarks based on video or audio encoding are also useless? Please clarify your statement.
 
i consider in game benches to be real-world, while things like superPI, 3dmark05, cpumark, etc are synthetic. there are some that kinda push the boundaries though
 
(cf)Eclipse said:
i consider in game benches to be real-world, while things like superPI, 3dmark05, cpumark, etc are synthetic. there are some that kinda push the boundaries though
Click to expand...
Also, running video and/or audio encoding would be real-world, such as running a 10-pass CCE on some DVD.

BTW, congrats on your custom title! :cool:
 
freeloader1969 said:
So the benchmarks built into games are totally useless then? Benchmarks based on video or audio encoding are also useless? Please clarify your statement.
Click to expand...

I did not say that benchmarks were useless, just that they are not accurate reflections of real world environments. In fact I thought I was fairly clear about where I feel benchmarks have utility - establishing relative performance levels and establishing, for a given price point, which equipment may be optimal.

Its a generalization, granted, but most real world setups run more than just a single application. This is true even of gaming rigs, especially if they are being used for multi-player fun. Its difficult and very time consuming to construct a benchmark that measures performance in such an environment. Consider that even a basic multi-player setup will have at least the game, net protocol stacks, firewall, and virus checker running. Constructing benchmarks that eliminate any impact from the last three is not something that is done often. In fact most benchmarks I've seen don't attempt to measure the behavior as affected by such things. I don't think that one can simply say the these programs will have the same effect on every system. I suspect, for example, that a single core AMD will have a different performance profile than a similarly priced hyper threaded single core Intel part when running in such a mix.

I'm not trying to make any judgements about a particular brand. Nor am I suggesting that benchmarks are useless or irrelevant. My sole point is that benchmarks are not measures of real world systems. They are, by their nature, measures of artificial environments. This does not mean that they are worthless. It just means, IMO, that they should not be used as the sole criteria for equipment selection.
 
He says "Sisoft Sandra is a very popular benchmarking program. Part of its draw is that users receive a performance rating for their computers within seconds. Since this is a synthetic benchmark, it sometimes yields results that seem out of touch with reality."

then goes out of his way to split the tests all to pieces making a whole chart out of any sub category intel manages to win......
 
freeloader1969 said:
What Athlon 64 chipset isn't stable? I've never ran into any problems with any A64 chipset, whether it be VIA, Nvidia or even ALI. They all run at stock speeds with no issues.
Click to expand...

The problem is that you can rarely pinpoint a single manufacturer or chipset to blame, because it's more complex than that. It's an obvious advantage that Intel can supply the chipset when they deliver the processor, as opposed to releasing a processor and being unable to control what your licensee's put to market. I've never had problems with my AMD motherboards either..... that I can blame on the chipset with certainty. But I'm always prepared to do the mandatory search for unbuggy drivers and hotfixes after I buy a new motherboard. Some of what I mean is described in this article. But I will admit that the truth in the quoted statement applies much more on the 'maturity' end than it does 'stability.' I hate to even write this, because I also cringe whenever someone brings up the usual "AMD chipset" argument. The reason is because it often comes out as "You can't have stability if you choose AMD" and that's obviously and absolutely not true.

I also wanted to state that my reason for posting in this thread was not to defend Intel by backing up THG's AMD bashings. PLEASE do not take this as my intent! I am reluctant when I post this because it may be interpretted that way. My point is that the relevant THG write-up is not completely tilted, and it makes fair concessions on either side of the argument.

Eclipse's "Clawhammer, Newcastle, Winchester or Venice" comment is a good example because, while the difference in these chips can be downplayed, it is a quirk that has presented itself as an annoyance on these very forums. How frustrating is it when a newbie tells us he has a 3500+ and asks a question, and we have to explain once again why '3500+' is not enough information? Does the stepping issue make or break AMD's business? No. But it is a quirk that should be conceded if you want to be able to move past petty shots in an AMD-Intel discussion.

The benchmarks do deservedly show AMD in a good light. But if THG wrote the meat of the article to parallel this and talk just about how l337 AMD was, then the article would be ignored most by the people who need to read and understand it.
 
RawsonDR...I'm no !!!!!! of either Intel or AMD. I buy what's appropiate to my needs at the time of a system purchase.

I hate to even write this, because I also cringe whenever someone brings up the usual "AMD chipset" argument. The reason is because it often comes out as "You can't have stability if you choose AMD" and that's obviously and absolutely not true.
Click to expand...

I totally agree with you. I can't even count how many times I've heard that old "AMD = no stability". It is too bad that AMD doesn't make it's own chipsets as you say. I guess their R&D resources are strained now between improving their current lineup and transitioning to 65 nm.

Eclipse's "Clawhammer, Newcastle, Winchester or Venice" comment is a good example because, while the difference in these chips can be downplayed, it is a quirk that has presented itself as an annoyance on these very forums. How frustrating is it when a newbie tells us he has a 3500+ and asks a question, and we have to explain once again why '3500+' is not enough information? Does the stepping issue make or break AMD's business? No. But it is a quirk that should be conceded if you want to be able to move past petty shots in an AMD-Intel discussion.
Click to expand...

Again, I totally agree. I've mentioned this myself in several threads :)

The benchmarks do deservedly show AMD in a good light. But if THG wrote the meat of the article to parallel this and talk just about how l337 AMD was, then the article would be ignored most by the people who need to read and understand it.
Click to expand...

That's another problem I have with the article. Tom's deliberately put the synthetic (Sisoft Sandra) benchmarks at the end of the article! Most people will only remember either the beginning or the end of an article. Most remembering the latter. This was done IMHO to put Intel in a "better light".
 
RawsonDR said:
Eclipse's "Clawhammer, Newcastle, Winchester or Venice" comment is a good example because, while the difference in these chips can be downplayed, it is a quirk that has presented itself as an annoyance on these very forums. How frustrating is it when a newbie tells us he has a 3500+ and asks a question, and we have to explain once again why '3500+' is not enough information? Does the stepping issue make or break AMD's business? No. But it is a quirk that should be conceded if you want to be able to move past petty shots in an AMD-Intel discussion.
Click to expand...
How's that really make it any different than Intel? I remember when people were clambering to find out if they had a Coppermine. Or what about Northwood? Presler? The fact that the P4 has increased the length of its pipelines over time? You can say that AMD did this or that, but Intel does the same stuff. The big difference is that AMD doesn't make you go out and buy a new motherboard when it adds features.

Besides... The vast majority of people actually buying these parts seperately actually know the differences, and if they don't, they're n00bs looking for an education, which they'll get from guys like us instead of bullshit biased reviews. If anybody's confused or annoyed, they should get educated instead of getting all emo about it.
 
RawsonDR said:
Eclipse's "Clawhammer, Newcastle, Winchester or Venice" comment is a good example because, while the difference in these chips can be downplayed, it is a quirk that has presented itself as an annoyance on these very forums. How frustrating is it when a newbie tells us he has a 3500+ and asks a question, and we have to explain once again why '3500+' is not enough information? Does the stepping issue make or break AMD's business? No. But it is a quirk that should be conceded if you want to be able to move past petty shots in an AMD-Intel discussion.
Click to expand...
this is quite true... but then again, the guys who don't tell us that often don't tell us a whole lot more info that could be useful too, so we end up asking anyway ;) :D
 
You must log in or register to reply here.
Back
Top