Sign in with
Sign up | Sign in
Your question

Bottlenecking Tests for 2009?

Last response: in Site Feedback
Share
a b à CPUs
July 17, 2009 9:36:45 AM

Hey guys,

This thread made me realize that there's no article (AFAIK) covering bottlenecks caused by low-end CPUs to more contemporary PCIe cards. Maybe something to revisit?

Regards,
r_manic
July 19, 2009 12:54:47 AM

Excellent idea, as well as a brief interpretation of bottleneck, and what it truly means/says. Theres been many a thread about how a mid ranged card is affected by a mid/lower range cpu.
I think an article like this maybe wont get ads, but itll certainly get readers, much like the P2 vs i7 article did
July 28, 2009 10:22:56 AM

Cleeve did one a while ago didn't he using old athlon xp's or something like that?

Still, it will most likely just confuse people and won't stop me ranting about it in an attempt to stop the flood of clueless parrots that seem to pervade on toms more than most tech sites.
Related resources
July 28, 2009 10:37:12 AM

Somehow, we really need to bottleneck those so called bottlenecking ideas
July 28, 2009 6:45:09 PM

Another PCIe lanes limitation through taping would be most welcome. This time for latest top-end singke GPU cards and PCIe 2.0 16x.
In fact they should do it for every new gen. of cards.

To BoM: Want more readers? Here's your chance. It's one of the few THG original review idea that's still selling like hot pancakes,
July 28, 2009 8:24:32 PM

Leave it to wuz for the right ideas. +1 wuzy. Great idea, and great article, and only makes sense, as each new gen of cards could pose a threat to any pci standard bandwidth, of which we have several now, unlike back then
July 28, 2009 9:48:02 PM

I'd love to see this.
July 30, 2009 12:04:24 AM

Those bandwidth articles are pathetic, they do nothing but do pointless tests without actually explaining the reasons why.

Look at any forum around the net and no one, no one actually can explain why usage would go up.

THe thinking is the more powerful the more usage without any backing up of why.

Show me an article which explain, without conjecture, just cold hard facts about what causes bandwidth usage.

People will do tests and all that and look at fps differences, usally average and try and deduce things from them but they show nothing, nothing at all of any use.
July 30, 2009 4:13:35 AM

Cleeve could do it. And since, as you say, no one else has done it satisfactorily, itd be a hit. Maybe have a blend in explanation, going easy, and also a lil in depth, whet everyones appetites, give the rookies a primer on it, while including the spicy stuff for the enthusiastically inclined
July 30, 2009 8:26:09 AM

I find it's always the conclusion that many authors can get caught out on.
It has to be a fine balance between making it as specific as possible and not being caught out by generalisation (ah yes, I don't forget easily). And making the above statement understandable by general public at the same time.

A large proportion of readers may skip straight to conclusion I believe.
Being a little different here what I do when I read a review of core components:
1. Scan read through the intros
2. Check test platform (hardware AND software) to make sure there's no unfair bias (e.g. if SSD is being tested with Windows XP, stop reading)
3. Start reading the data and look for real-world usage ones, especially relating to your work. Disregard all raw synthetic benchmarks unless I'm researching on academic basis.
4. Read the conclusion of the author. If it matches mine then he/she gets a big tick in my book (very hard to come by btw). Partially matching, ok a pass. Totally biased and unfair, BLACKLISTED!
July 30, 2009 9:22:11 AM

^^ Agreed. I think synthetics are used because they may be popular, but often pointless, matching real world apps, huge bonus, apps in same family/type differ too wildly and if set up isnt as close as can be made, without a mention or caveat to any changes, spells the reviewer doesnt know his stuff, and isnt at least following thru, as some sites are smaller, and simply dont always have on hand whats needed, which may or may not cloud the final verdict
a b à CPUs
July 30, 2009 10:44:07 AM

Only reason why synthetics are pop is because they are easy to test with.

For me, I just skip them and look at program I actually use :) .
July 30, 2009 11:47:31 AM

True, using a synth gives you tons of data, all worthless, but it fills pages heheh
a b à CPUs
July 30, 2009 12:06:43 PM

no, no, you get AMD fanbois to do that for you.

Their FUD is much more interesting to scoff at.
July 30, 2009 12:30:45 PM

But there does exist the rare but often seen AMD fangirls about
a b à CPUs
July 30, 2009 1:05:55 PM

yes...

new avatar now... what do you think?

July 30, 2009 1:26:09 PM

Like it! Nicer smile, looks cheerier, plus, you can tell youre getting better
July 30, 2009 1:32:01 PM

Imagine what would happen if jaydee's avatar and your one produced off spring, scary thoughts.
a b à CPUs
July 30, 2009 1:43:53 PM

JAYDEEJOHN said:
Like it! Nicer smile, looks cheerier, plus, you can tell youre getting better


I can do alot better but I like it in that cartoonish style :) 
July 30, 2009 1:50:41 PM

Nuthin but a bunch a red headed step chiles heheh
a b à CPUs
July 31, 2009 5:09:51 AM

There is an article coming up that touches on PCIe scaling with CrossfireX but it's not specifically on the topic so the data might not mean much to you. It's going to hopefully get rid of some of those "Gaming PC, buy now or wait for i5?" threads.
!