Tomshardware and their flawed benchmarking methods

xoleras

2[H]4U
Joined
Oct 11, 2011
Messages
3,551
I read this article and busted out laughing:

http://www.tomshardware.com/reviews/graphics-card-benchmarks-charts-review,3154.html

Apparently Tomshardware is using override Edge AA on AMD cards and not using a corresponding setting on nvidia cards. Yes, we all know the GTX 680 is faster but this would explain how lopsided their results are: The reviewer is an idiot and used Override AA on the AMD card.

Override AA is slower than native ingame AA and apples to apples comparisons should obviously be application preference across the board. Maybe someone should clue him in?
 
...once again proving HardOCP is one of the best tech sites for GPU reviews. They actually know what they are doing.
 
You would not believe how many people I know that swear by Toms..
Its pathetic.
 
You would not believe how many people I know that swear by Toms..
Its pathetic.

Toms has been around forever and ppl just stay with what they know. I like TPU and this site best but i got into game late. Since i go here first usually I could see myself missing other newer places.
 
I like Toms. Its not my favorite though and I use [H] and Anand as the gospel but I still enjoy Tom's articles from time to time. Most of the time they jibe with what I read here and other places but in the event they don't, I'll just defer to [H] as the correct answer.
 
You would not believe how many people I know that swear by Toms..
Its pathetic.

I haven't touched that site in ages. Hopefully they've improved a lot, but respect is hard to earn back. TPU is nice but I wish they'd remove stuff like 1024x768 and upgrade their Nehalem for their 7990/690 review. :D
 
Other than their sometimes bad and or biased reviews my biggest problem with Tom's today is the 18+ scripts active on their site.
 
I look at all of them.

this is the wisest option. never healthy to get all your news (tech, hardware enthusiast or any news for that matter) from one source. having said that, i always have the [H] mainpage open in my browser (often the first page i load when starting a new session) and several tabs for the [H] forums.

my usual go-to tech/forums are here, ocau, ocn, anand (not the forums tho) and tom's.
when tom sold off the site, started losing interest there. not real keen on their page layout either. also the forums over there are just bad. some of the advice i've seen given over there is just flat-out incorrect. it's quite worrisome actually.
 
[H] and AT write the best reviews. I've also found that their results are the closest to what I get when I actually use the card, and after all that's what I want to glean from reading reviews anyway.
 
I read this article and busted out laughing:

http://www.tomshardware.com/reviews/graphics-card-benchmarks-charts-review,3154.html

Apparently Tomshardware is using override Edge AA on AMD cards and not using a corresponding setting on nvidia cards. Yes, we all know the GTX 680 is faster but this would explain how lopsided their results are: The reviewer is an idiot and used Override AA on the AMD card.

Override AA is slower than native ingame AA and apples to apples comparisons should obviously be application preference across the board. Maybe someone should clue him in?

Where exactly does it say that? Please point it out.

Edit:
I had a recent conversation with Igor and he told me that THG benches with current drivers for ALL cards and with currend game versions. This is not self-evident (although it should be)! Also I think it is commendable to explain the benchmark methodology - again, not everyone does that.

Edit 2:
I think the OP started his rant on his own confusion. Look what we have here...edge AA. This has nothing to do with AMDs edge AA whatsoever. Nowhere in the whole GTX680 review (or any other recent review for that matter) does it say "override" or "AMDs edge AA".
 
Last edited:
They're CPU reviews are aight, but their GPU reviews are pretty worthless there...
 
Everyone is biased one way or the other.

The [H] wasn't always perfect at one time either......

Thankfully, people learn from their mistakes, and the [H] is now one of the best review sites to date.
 
I've had a beef with the way Tom's handles their site and articles in particular, though not necessarily their reviews. I've found their reviews can be very hit or miss, with some great reviews that go to lengths others don't or they stop short of actually providing enough info (efficiency charts for CPUs are great vs poor GPU reviews). The thing that bugs me the most is that the site has turned into ad space for hardware vendors where news articles read like they're directly from the marketing/PR guys' mouths. If I have to click through a popup to get to your site only to read an article that tells me nothing other than a new product that nobody gives a shit about has been released then I'm likely not going to visit very often.

It's a shame too. They tend to have some awesome articles like the recent openCL/directcompute ones that I hadn't seen anyone else do but when they're so concerned with ad revenue and not the integrity of their journalism and reviews it's no wonder they've fallen from grace.

The sites I do frequent the most are AT, [H], TR, ars, extremetech and xbit while the others get a glance or two a week.
 
Tom Pabst is the antichrist.

Just thought you should know...
 
He's pretty boring compared to when he was on the Intel dole back in the day. Anyone remember the beef between Tom and the guy who ran Van's Hardware?
 
I read as many different articles as I can including Tom's because there are always bias's everywhere, intended or not. I get to see way more games/programs benched that way. I think they try to do honest reviews but sometimes their methodology is out of whack. The article comments gives you a pretty good idea how people react to their testing methods, and they do get called out on it from time to time.
 
He's pretty boring compared to when he was on the Intel dole back in the day. Anyone remember the beef between Tom and the guy who ran Van's Hardware?

Tom has has a beef with just about everyone at some point, which is why he's generally regarded as a colossal prick.

Most of the old timers at the [H] look at TH articles only for the chuckles, mainly because if you look hard enough you will always find something flawed if not monumentally stupid. What you young folks would throw a <facepalm> at... :D
 
I read this article and busted out laughing:

http://www.tomshardware.com/reviews/graphics-card-benchmarks-charts-review,3154.html

Apparently Tomshardware is using override Edge AA on AMD cards and not using a corresponding setting on nvidia cards. Yes, we all know the GTX 680 is faster but this would explain how lopsided their results are: The reviewer is an idiot and used Override AA on the AMD card.

Override AA is slower than native ingame AA and apples to apples comparisons should obviously be application preference across the board. Maybe someone should clue him in?
Yeah, I'm a relative Newbie to GPU overclocking, benching and even I knew that! :rolleyes:
 
Unless the OP actually clarifies how he came to this conclusion (no proof was presented yet), this thread is full of BS.
 
Tom's = sad.

Anyone that buys there review methodology needs to rethink there position.

When it comes to video card reviews its [H] and [H] alone. Call me biased , I don't give a shit but I think this site does it best.
 
So, is the OP going to clarify his position?

Also, tom's hasn't been owned or run by the original guy for how many years?

I don't think you can find a hardware site that isn't ridiculed by others so this sort of thing is backward to say the least.
 
lol i didn't even make it for enough in tom's arcticle for the 7970/680 to even notice the numbers were wrong. their articles are rubbish, i don't care for anandtech's testing either but when it comes to technical articles like explaining how architectures of gpu's/cpu's and what not work they are the go to people. they may not know how to test things but they sure as hell know how to write in-depth technology articles.

i prefer [H] for my GPU reviews but i'll also go look at other review sites that use my resolution in their testing(1920x1200) when [H] doesn't test that resolution in their review just to get a best case scenario for performance at my resolution.
 
...once again proving HardOCP is one of the best tech sites for GPU reviews. They actually know what they are doing.

While I enjoy this website for reviews; for technical information and other data I tend to prefer Anandtech. I'm not a fan of the way this website collects data for benchmarks. I wish they'd just do the 'max playable' tests like they currently do, but also supplied a clean and easy to compare graph like Anandtech / Guru3D do.
 
Read Tom's Way back in the day....

Not anymore recently. Just going to the website alone hurts my brain. I always try and remember to use [H]'ds links when I buy stuff in support hoping this site never looks like Toms....
 
I find it best to check as many review sites as I can to get a good over all feeling of how the product is represented. Toms benches a few games that others do not, and some of them I play so I do check his out most of the time, along with [H] and a few others that I think do a great job!
 
Where exactly does it say that? Please point it out.

Edit:
I had a recent conversation with Igor and he told me that THG benches with current drivers for ALL cards and with currend game versions. This is not self-evident (although it should be)! Also I think it is commendable to explain the benchmark methodology - again, not everyone does that.

Edit 2:
I think the OP started his rant on his own confusion. Look what we have here...edge AA. This has nothing to do with AMDs edge AA whatsoever. Nowhere in the whole GTX680 review (or any other recent review for that matter) does it say "override" or "AMDs edge AA".

Since the OP has not redacted his claim, I can only assume he was spreading FUD.
 
While I enjoy this website for reviews; for technical information and other data I tend to prefer Anandtech. I'm not a fan of the way this website collects data for benchmarks. I wish they'd just do the 'max playable' tests like they currently do, but also supplied a clean and easy to compare graph like Anandtech / Guru3D do.

You have a hard time with Apples to Apples tests with numbers on those? Do you have to have a bar?
 
Back
Top