Sign in with
Sign up | Sign in
Your question

THG/Intel Conspiracy Theorists: Post your proof here!

Last response: in CPUs
Share
January 13, 2006 3:46:11 AM

First off let me begin with saying the post is not inteded to be flame bait (I have no doubt that it will however become flame bait). I'm genuinely interested in seeing any evidence anyone has to offer supporting the theory that THG is biased towards Intel.

Today I was reading a thread and someone said something to the effect that THG's CPU benchmarks cannot be trusted and are Intel biased. To date, I have not seen a single shred of evidence to support this claim. Admittedly however I have not looked very hard. But whenever I confront a THG conspiracy theorist about proof, they often say "well look at HardOCP and Anandtech.com, they're benches are different." This is not proof that THG is biased. And one could argue that those two sites are AMD fanboy sites (especially HardOCP). To date no conspiracy theorist has shown me anything even remotely difinitive on this topic.

So conspiracy theorists, here's your chance to prove your theories. Please post evidence showing that THG is biased towards Intel. I've also heard many theorists claim that THG takes kickbacks from Intel. So if ya'll have an internal memo, definitive article, or anything like that, post it here.

Remember, I want EVIDENCE; not speculation, theories, rumors, or flame posts.

-mpjesse
a b à CPUs
January 13, 2006 3:57:45 AM

"Don't ever believe THG benches. They have to be Intel approved."

The remark that lead to this topic. The funny thing is I've been visiting THG since 2000 and I have yet to see ANY benchmark that proves that remark. If you look at the benchmarks for the last year, Intel has been losing to AMD big time in many key areas. So, if the THG benchmarks are Intel approved, why do they hurt Intel then?
January 13, 2006 4:32:41 AM

I've never heard this issue before, why do you people think that THG is unfair towards INTEL?
Related resources
Can't find your answer ? Ask !
January 13, 2006 4:47:37 AM

I personally do not believe it to be true, however about 25% of the people on these forums and others believe it. Just go over to anandtech's forums or HardOCP's forums... all they do is trash talk THG and talk about how biased they are.

It's even worse in here sometimes...

-mpjesse
January 13, 2006 4:52:37 AM

To start with, let's look at the most recent cpu review, the FX60. 3 games, and 2 are quake.
Oh sure, there are other games, but who understands those graphs?, and can use them to compare?
Why are they using beta tests?
Wonder why they stopped using office tests?
Benchmarks seem rather heavy in favour of encoding tasks. Mostly ones that favor intel
Futermark- oh yah, i believe the P4EE 3.4 is 10% better than the FX 57
January 13, 2006 5:12:21 AM

In recent times I would have to say its rather the opposite. Mostly because AMD is winning so much in the benchmarks. Tom's even goes out of its way in the conclusion of a recent artilce to say Intel only won 8 of the benchmarks out of 32 in their recent "Mother of all CPU charts" (i think) with 7 being senthetic and only 1 being a real world test.

disclaimer: some of the facts maybe off a little but they should b close :) 
January 13, 2006 5:39:18 AM

I do not consider this answer to be proof, but let's look into it further:

Quote:
To start with, let's look at the most recent cpu review, the FX60. 3 games, and 2 are quake.


I agree they should have benched more games than just Quake 3, 4, and FEAR. But as for the graphics- Quake 3/4 does not have the capability of measuring minimum, average, and max frame rates like FEAR does. (This can be done w/ FRAPS however). At any rate, I'd rather see benches w/ minimum, average, and max frame rates.

Now, I did do a little research. THG benched Quake 4 as being faster on an Intel 955 system while Anandtech benched Quake 4 being faster on an AMD system. Compare for example Anandtech's benches to Tomshardware's in Quake 4. (both sites used the 1.05 beta SMP patch)

http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=266...
http://www.tomshardware.com/2006/01/10/amd_athlon_fx_60...

Anandtech has Quake 4 being faster on the FX-60 while THG has Quake 4 being faster on the Pentium 955. But if you take the time to look at the test setup, you'll quickly figure out why. Anandtech is using DDR2-667 timed at 5-5-5 while THG is using DDR2-667 timed at 4-4-4. That's a huge difference. And while I'm on the topic, why in god's name is Anandtech using 5-5-5 timings on their Pentium 4 testbed? That hardly seems fair since they're using premium OCZ RAM timed at 2-2-2 on their AMD testbed. If fast timings are good for AMD then they should be good for Intel too. (and it ain't like 4-4-4 timed DDR2 is that hard to find... it ain't that expensive either). Now that's what I call test rigging in AMD's favor.

Quote:
Why are they using beta tests?


All hardware reviewing sites are using the beta patch for SMP in Quake 4. If you were refering to the xvid beta codec, everyone i know is using that codec to encode.

Quote:
Benchmarks seem rather heavy in favour of encoding tasks. Mostly ones that favor intel


Like it or not, a lot of ppl encode video. Xvid, WMP, and Divx are the most popular. Compare the benchmarks shown here for WMV9:

http://www.tomshardware.com/2006/01/10/amd_athlon_fx_60...
http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=266...

They both got the same results. I can't compare any other encoding programs because Anandtech only tested DVD Shrink and Windows Media for encoding.

Quote:
Wonder why they stopped using office tests?


Now you might be on to something there. They haven't been using office benchmarks- you're right. And AMD wins office benches hands down. This is the first logical example of possible bias.

Quote:
Futermark- oh yah, i believe the P4EE 3.4 is 10% better than the FX 57


Unfortunately Xbit Labs and Anandtech chose not to benchmark the FX-60 using 3DMark05 CPU test. But firingsquad.com did. And they got the same results THG did... See here:

http://www.firingsquad.com/hardware/amd_athlon_64_fx-60...

So... I still don't see any diffinitive proof that THG is biased. What I see is a lot of different system configurations among review sites that lead to different scores based on a number of things such as RAM timings, benchmark apps, etc. The only interesting thing you mentioned was THG isn't using office benchmarks. I guess that <could> be construed at bias. However, I don't see THG's rigging tests like Anandtech is.

-mpjesse
a b à CPUs
January 13, 2006 5:46:03 AM

Actually, THG is not bias in my opinion. They are just saying the same thing many reliable sources are saying: AMD is a better chip overall.

The reason they don't use Office for benchmarking is simple:
Office really doesn't push today's processors anyway! Word 2000 is the same speed on my 700Mhz Pentium 3 as it is on my 2.5Ghz Celeron. That's why!
January 13, 2006 10:24:10 AM

I prefer my overcocking to be with females... :tongue:
January 13, 2006 2:12:30 PM

Lol, that stress test thread got way out of hand. No one could believe what they were seeing...and the conclusion to the test. Wow. That was bias.

But saything that, since the revamped site I've found they have becoming increasingly fair with their conclusions and articals concerning AMD chips.
January 13, 2006 2:43:03 PM

Quote:
But saything that, since the revamped site I've found they have becoming increasingly fair with their conclusions and articals concerning AMD chips.
:lol:  :lol:  :lol: 

I wouldn't know about that. I haven't read their articles in forever. They were too biased and/or too much just crap. Maybe they are better now. **shrug** But without anything actually interesting going on in the PC industry at the moment, who really cares? :lol: 

Besides, the new site layout is so bad I doubt I'd be able to find anything even if I did look.

What I really miss are those good overclocking articles. Not what to do with a tank of LN2 and a lot of spare time, but, you know, home achievable regular-use stuff.
January 13, 2006 2:48:55 PM

Quote:
The reason they don't use Office for benchmarking is simple:
Office really doesn't push today's processors anyway! Word 2000 is the same speed on my 700Mhz Pentium 3 as it is on my 2.5Ghz Celeron. That's why!
Anyone who believes this doesn't actually use Office. I mean really use it.

I'm sorry, but it's true.

I write novels, mostly in Wordpad because it's lightweight and doesn't slow me down. Then when the rough draft is done I fix things up in Word. Let me tell you, you'll notice a difference in performance with different procs when Word starts running a 400 page novel through a spell/grammar check in the background to highlight all of the mistakes. Hell, you'll notice a difference in platforms just opening a document of that size.
January 13, 2006 2:49:35 PM

Quote:
What I really miss are those good overclocking articles. Not what to do with a tank of LN2 and a lot of spare time, but, you know, home achievable regular-use stuff.
Bah! Who needs that when you've got US! 8O :lol: 
January 13, 2006 2:56:10 PM

Quote:
Bah! Who needs that when you've got US! 8O :lol: 
:lol:  There is that, but I just enjoyed reading the articles. I don't actually have the money to do these things. :( 
January 13, 2006 3:14:58 PM

I agree... I use excel a lot for work. Opening a 4MB workbook on my shiny new 3Ghz Prescott Lenovo system is much faster than it was on my old P4 2.4Ghz Dell.

(note: these are both work computers, I did not build/buy them)

-mpjesse
January 13, 2006 3:31:06 PM

Yeah, Excel is another one that can really eat things up.

Powerpoint and Outlook, at least, are pretty much just memory-based. They can be real memory hogs though.
January 13, 2006 4:33:42 PM

Do you have a link to that article? I can't seem to find it. And how is it biased?

-mpjesse
January 13, 2006 4:47:47 PM

BTW, I just email Anand about his recent FX-60 article (which can be seen here: http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=266...)

This is what I emailed him:

Quote:
Anand,

Thanks for taking the time to review the FX-60. A very nice processor indeed.

However I have a bone to pick with you about your test bed. In the AMD test bed you used OCZ PC3500 memory with 2-2-2-7 timings. Those are nearly the lowest possible timings you can get with DDR SDRAM. In the Pentium 4 test bed you used DDR2-667 Memory (you didn’t mention what brand or how much) with timings at 5-5-5-15. Those are the highest possible timings you can find w/ DDR2-667.

So my issue is this: if ultra low timings are good for AMD then they should be good for Intel. Did you know I could only find 3 DDR2-667 modules on pricewatch.com with those timings? And NONE of the memory was name brand. In fact, OCZ, Corsair, Geil, and Kingston don’t even make DDR2-667 DIMMs at those 5-5-5 timings. Only really, really cheap generic manufacturers make DDR2-667 DIMMs at those timings.

It seems like you went out of your way to make sure Intel had the crappiest memory available. I wonder why that is?

Jesse Sollien


-mpjesse
January 13, 2006 5:48:22 PM

While TwitchGuru IS part of THG network, it's not THG persay. However that article is all farked up- ur right. It should be burned at the stake.

-mpjesse
January 13, 2006 5:53:47 PM

The way I look at it, every review site is biased. How easily you can identify that bias and balance it against other reviews is what matters. So long as the review site lists full details on their test platforms so that I can see where their bias is, I'm good. :lol:  :lol:  :lol: 
January 13, 2006 9:10:56 PM

I think maybe wusy was refering to this comment specifically:

Quote:
Stability: Intel With Intel, AMD With NVIDIA

An important topic, particularly in the professional arena, is system stability, and there were a few surprises in store during the stress test. In the Intel system a total of three boards with NVIDIA's nForce4 SLI chipset caused problems. Outages were frequent, and in one case the voltage regulator even blew. Help came in the form of a combination of an Intel CPU with Intel chipset. However, the integrated SLI graphics setup is then unusable.


Everyone knows that nVidia works best with AMD. When nVidia designs flagship motherboard chipsets, they always start with AMD. So it should be no surprise that Intel doesn't run very well on an nVidia/Intel based SLI motherboard. That's a bit like trying to run an AMD Athlon 64 on an intel based motherboard (yes I know it's impossible, but what if it were? the two certainly wouldn't run well together)

And then there's this statement:

Quote:
The AMD platform offers just one choice, which is also very stable: the Athlon 64 X2 4800+ with a board featuring the nForce4 SLI chipset from NVIDIA. The results of our stress test show that the nForce4 SLI chipset is better suited to the AMD platform.


This is one of the most retarded things I've ever heard. I guess THG isn't aware that ULi, SiS, and AMD itself makes chipsets. AMD would be owners have just as many choices in chipsets as Intel would be owners do. In fact it is Intel who a lesser number of chioces based on the fact that nVidia's board didn't work so well with the Pentium 4.

So now we have another possible legitimate link to the bias. While the person who wrote this article is obviously a moron and probably biased, it does not prove THG itself is biased. And it certainly doesn't prove that THG skews it's benchmarks in favor of Intel (which is the common perception of some ppl here).

-mpjesse

P.S. Thanks wusy. :-)
January 13, 2006 9:12:34 PM

Quote:
Let me tell you, you'll notice a difference in performance with different procs when Word starts running a 400 page novel through a spell/grammar check in the background to highlight all of the mistakes.


8O you really make that much mistake!! :wink:
January 13, 2006 11:11:55 PM

Quote:
One thing for certain, the Cheif Editor(what was his name again? David something?) for THG is biased. There's a few articles written by him which I'm trying to find now.

BTW, you're welcome. Just doing my job.


Was it David Hasselhof?
January 14, 2006 4:19:47 AM

I dragged up the stresstest 11 post, just for you.
Quote:
I agree they should have benched more games than just Quake 3, 4,

The reason they use quack is because it uses SSE for the Intel, but will only allow the Amd to use 3D now. There is an Amd patch, which averages a 20% increase in framerates.
Quote:
Like it or not, a lot of ppl encode video.

Yes, but how many use the high priced progs listed. BTW, I suspect that the listed encoders were built using intel compilers, just as most Adobe progs are.
While a lot of people do encode, it seems that there are a lot of encoding benches.
Quote:
Unfortunately Xbit Labs and Anandtech chose not to benchmark the FX-60 using 3DMark05 CPU test. But firingsquad.com did. And they got the same results THG did...

I'm not questioning the result so much as the test itself. Look at the results and tell me you think they are a good and true representation of chip performance. This is a synthetic benchmark, so it would be nice if it somehow reflected reality, dont you think?
In a real world, a test sight would look at the results, laugh and send the benchmark back, with a note questioning the sanity of the maker.
January 14, 2006 12:40:16 PM

Put a Formula1 car on a track against..let say.. a KIA Rio. There is no doubt the F1 wil win. this is benchmarking.

Put the same Formula1 and Rio in city drive with lot of traffic jam and light. But car will go from A to B in about the same time. This is real life uses..

I don't care about benchmark. Especially syntetics benchmark. They means nothing to me. Just like comparing video card with sound off and low resolution. If I get a x1800xt, I don't expect to play with 640x480 and no sound. Better test would be with full sound on and the measure, should be image quality and fluidity. Did the card able to display the best image possible while running with sound at best too. Because it is mostly what I'll be doing.

When I see benchmar, from any web site, I take them with care. First, having a direct comparison between CPU brand is likely impossible. Too many factor are going to influence data. Motherboard, RAM,drivers, .. all of those can make a CPU looks good or bad. To have a fair comparaison of hardware, DOS based benchmark should be used.. no chipset drivers, no hdd drivers... just raw x86 code, no sse or amd 3dnow. those should be tested alone. Then benchmark should be written directly in assembler. No compiler optimization, only human knowledge. Just do something simple.

Then, as for THg being biased, well, IMHO, they are not always fair in their game. The stress test was a nice exemple. They lookd like unprofessional when they did that. What turn out to be a stress test, to stress material rapidly turn into a benchmark test to negate the problem they had at first with Intel system. Sur benchmark here are a nice addon, but they is some questions about reliability that should have been discussed with the maker of the board that failed and not only with Intel why their CPU is burning board (answer was to use Intel mobo..). Does the board maker has full specs of this CPU to design a board able to support it or if Intel just throw there a more powerful CPU to counteract the AMD in a rush and forget to tell others about new power requirement but themself? A think of a lot more..

Seriously, this test should have be about why Intel manage to kill all the motherboard but the Intel one and why others has failed.. It is a stress test .. not why this cpu was faster than this one in this case.. I don't care.. I wanted to know about reliability. With the amount of power drawn by system, how do weaker PSU would have done here, because wel all know that people don't only buy good PSU.. What is more likely to be affected by heat... No, they could not do that.. It was all Intel that was giving them problem.. That would not have looked good.. But they for sure point out that the AMD system would not turn off. wanted to stay on.. Maybe it is smarter that them.. knowing what stress test is? running and running...
January 16, 2006 12:32:54 PM

Quote:
Let me tell you, you'll notice a difference in performance with different procs when Word starts running a 400 page novel through a spell/grammar check in the background to highlight all of the mistakes.


8O you really make that much mistake!! :wink:D amn skippy I do! :mrgreen: I spell (and type) like a drunken sailor at times. Huked on fonix wurked four me! Though I'm getting better, a lot of my novels have been worked on at various time over years, so there's a lot of bad old stuff in them. He he he. And it doesn't help that the replace-all feature in the spell checker only seems to go forward for so many pages and not do the obvious thing of replacing all.

Though slightly more seriously, Word's grammar check is also about as accurate as my spelling typically is. (At least Word 97's is.) So it marks a lot of things wrong that aren't, and still misses a number that are wrong on top of that. :roll: So running that check alone can suck up a lot of processing power.

Quote:
Seriously, this test should have be about why Intel manage to kill all the motherboard but the Intel one and why others has failed.
:lol:  I got the feeling this was mostly because motherboard manufacturers didn't quite believe Intel's ambition to reach the maximum of their power specs. :wink: At least at first. I'd hope that they've figured it out by now...
!