I miss a benchmark for filtering/sorting although I'm sure it's in a small percentage part of the tests I'd like to know specifically how a CPU does in sorting/filtering/regex stuff since I find I do use that plenty.
Also I find the choice of winzip, which nobody uses anymore, an odd one, why not 7zip or winrar? Winrar is most intensively used in the real world, and 7zip has a nice inbuilt benchmark making life easy for reviewers.

Oh incidentally, the filtering, sorting and RegEx stuff would also help indicating how fast a CPU can do a virusscan, which is another thing a CPU can be busy with for some time.

I love the article, those graphs are awesome and really put things into perspective.

Couple of comments.
Do the same thing for video cards sometime (obviously with the same CPU though)
Also, can you include overclocking, PLEASE on some of these?

I'm / We're (?) fine with taking it with a grain of salt and yes we know X clock speed is not guarunteed but just one page with 2 games and a solid Windows benchmark would be nice.

I mean the AMD's may well go MUCH better with overclocking, the Intels not as much and the i7 920 might well be amazing or terrible who knows

This is an enthusiast site, overclocking plz :(
Thanks forthe good article.

P.S Anyone else own a dual core E8XXX series and just can't find any damn reason to upgrade?

Yeah even if you do some kind of "easy overclock point" it would be totally so much more useful.

I am doing some research to look into what CPU to buy for my gaming PC, including overclocking in gaming benchmarks is just so important I can easily afford a i7 Core 2, but if its not going to be any faster under my preferred games at my preferred resolution then why bother? I may as well go for a faster GPU.

I hit this site/review with a big amount of excitement only to find it couldn't be any more useless to my research its just same same old rubbish I can find anywhere else.

Its not hard to do, just put OC'ed next to the correct graph.

Hey guys, can you throw in a Q6600 overclocked to 3.4 GHz and compare it to the rest? To me, this article is useless without overclocked CPU's.

Just thoughts :)

I feel that overclocking should NOT be added in these reviews. The main reason being that its unreliable. Every chip has a different overclocking point, and putting that into a dollars/fps graph would be just plain stupid. If you say you can get BLANK chip for a certain price and it can OC to something, than someone buys is (that isn't a huge enthusiast) and it can't OC to w/e TR had it at, they have been mislead. I feel TR is one of the only review sites that doesn't mislead its readers, and it has no reason to start that now.

Though I do feel that a graphics card version of one of these reviews could be a great help to many users. And I feel that you covering potential last gen-cards (similar to how you covered the X2) is important, as you can pick up great deals on 8800GT's and HD3870's for below $60 if your lucky. For prices like that, sometimes they are worth it.

Awesome review, nothing short of TR's top 10 best IMO.

Nice article.
But hm...
why don't use more CPU-intensive game like GTA IV for benchmarking?

*pets his overclocked e8400 and purrs contentedly*

Awesome article TR. I read every page and dugg it too.

The best part was page 11 when you configured 3 systems and their cost and plotted that too.

You've saved me hours of research when it comes to system building and buying decisions.

Thank you.

good info to know, havent been keeping up with hardware lately but this keeps me in the loop a little

thanks

I'm amazed at how perfectly linear AMD's performance/price comes out over all their chips most of the time. I'm impressed.

Looks like I was right to get my E8400 a few months back :) It hasn't really dropped in price since then, and it has some ridiculous headroom for overclocking.

While this situation is definitely easy on the wallet, I'm beginning to wish some game devs would push the envelope a bit more. We haven't even seen anything match Crysis, let alone surpass it (and that came out 1 1/2 years ago now)

So in conclusion, let's all buy Athlon64 X2 6400+ chips.

:P

Cue the usual bitching about game benchmarks running at "unrealistic" resolutions. (It would be amusing to see a graph with all the markers clustered tighter than the WorldBench Office one, though). Meanwhile, as someone who skipped right to the non-gaming benchmarks, I'd like to thanks for all the work you put into this.

OK, so in Firefox 3.0.7, if you're using the extra-wide layout, the table in the article still isn't as wide as the article, so you still have to scroll (there's a big empty space on either side of the frame, though)

Yet again, FPS-itis strikes. So, you pick a bunch of games from the LEAST CPU intensive gaming genre and make broad declarations. This is getting tiring. Can NO site benchmark a CPU for gaming? Where's the RTS games? Where's the flight sims? Yet again, FPS is the only genre of games ignorantly benchmarked for a CPU test. YOu pick the ONE genre known for GPU limitation and not much CPU interaction and showcase it. This is ridiculous.

You know what would make those scatter plots easier to read? Diagonal lines going from the lower left to the upper right. Maybe put them in a different color or something. I do not know if that is possible but it would be helpful.

How about the Atom 330?
I know it's a different class of cpu but it made the "Kitchen PC" and when the Intel motherboard AND cpu cost $79.99 the value proposition starts to look good for applications like, well, the kitchen pc.

Great article as usual.

I too think it's a shame that no e7000 series CPUs were included especially since they ranked so high in the previous article in this series. That's my only real complaint.

PhenomII 920 isn't that bad. Shame it's not an AM3 chip.

Wow, you guys outdid yourselves this time. Not only did you include a bar graph of "frames per dollar" metric, but you even had a "systems cost-added" scatterplot. I don't know if either of those were requested, but I liked them a lot.

The 7750 is miserable. I just ordered a $55 X2 5800+ -- I've been using a $20 overclocked Sempron LE-1250. My E7200 can run at 4.2GHz in a $55 motherboard--that would score well in the performance/$ category, but it does use a lot of power.

These are the coolest articles. Thanks Cyril.

Once again I'm disappointed in your choice of processors today.

* No E8500? It's $20 more than an E8400 and slightly faster.

* You included the E8400 (3 GHz) and the Q8200 (budget quad core) but not the E7500 (budget dual core, 2.93 GHz), which is nearly as fast as the E8400 and $30 cheaper, with the exception that it lacks VT and has half the L2 cache (but still a copious amount). I suspect the E7500 would do substantially better than the E8400 on a bang-for-buck basis.

* You included an Athlon 64 X2 6400+, which is out of production for the retail market, but not the newer, K10-based 2.7 GHz Athlon X2 7750, which might have been more interesting. And at 95 W it's even rated for a lower TDP than the 90 nm, 125 W 6400+.

* If you follow the slope of a line joining the origin with the datapoints in your overall system scores, the Q9550 more or less comes out on top if we maximize performance within the highest range of bang-for-buck.

For your average graph at the end, why not make the prices for everything be a percentage over what the 6400+ cost is? That way both axis are percentages.

My Q9300 looked so much better last time.

Perhaps Meadows should have pushed for his unit more in this review.

For productivity tasks time is money. I wonder how rates of $10/hr and $50/hr would affect the value proposition on productivity tasks?

Boy, that NCQ issue on the AMD platform really hurts them in a couple of the tests. But I thought some of the folks here reported getting NCQ to work on the 700 or 750 SB. Is that incorrect?

Might it be worthwhile to plot the price/performance curve, and note significant deviations from the curve.This would identify parts that perform much better (or much worse) than typical. I think it's generally understood, and well illustrated by the logarithmic trend in the price/performance charts, that significant premiums at the high end will result in nominal increases in performance.

I guess what this means to me is that my trusty E8400 is good enough for now. The only thing I need is a faster video card.

Very interesting and thorough article. Exactly why I read the Techreport.

However, I think it's pretty bad news for CPU and PC manufacturers.

The raw performance difference between the top and the bottom is about 3X, but this just doesn't translate into the most used real world apps.

I consider myself a PC enthusiast, but I don't game and only very occasionally encode a video (hey if I can bittorent a 40minute show in 7mins, why bother? No i7 will do that..). There simply isn't the need to upgrade anything from the dual-core 64-bit era (what 3+ years now) for 90%+ of the PC using community. RAM is almost being given away. Stuff any old machine with 4GB worth, ramdisk browser and media player caches and you have a rocket-fast PC for nearly every task.

I've been using netbooks as my primary PC for 6 months now. My most used desktop is a 690G machine by virtue that it has HDMI for my TV.

Power is great when you need it, it's just that most of us don't, at least not for most of the time. Portability and connectivity are far more important. The biggest limitation is in front of the keyboard - how fast you type and how quickly you can read. I'm not in the bit surprised notebooks have overtaken desktops in sales and the only overall increase in sales are in the netbook area.

Damn - I wish I had an excuse to build a new machine, but I just don't lust after the latest and greatest anyone.