Should benchmarks be run with processors only running stock speeds?
I post this as we all know the true potential of Intels overclocking. Joe Average walks into a computer shop to buy a processor is going to buy it at stock speeds.
With the headroom Intel has for overclocking while AMD does not necessarily, its only fair to benchmark and post benchmarks at stock speeds right?
If you look at the benchmarks, most people say they are biased. Its really hard to say how much of this is true. But, if you look at every single benchmark of the FX60 vs. EE, you'll see at most points that the FX60 is the clear winner. Most people are not going to overclock a processor and why should one have to overclock a processor to beat its rival by a narrow margin?
I know this may seem like a dumb thread but, I gotta ask this question as I think its only fair and the point is once again, we buy processors at stock speeds. We can't go buy an Intel processor clock at 5Ghz or likewise, an AMD.
Well, if the bios/mb doesn't allow OC.. err.. ya. More then likely an average Joe would purchase a brand based computer, and go by the salesmans advice on what would be best to buy.
I don't think the average Joe would look at benchmarks to make that decision. Unless he stayed at 'Holiday Inn' and feels smarter, and starts to ask questions without really knowing what he's trying to ask. :wink:
But is it fair to compare stock speeds, but also take in consideration on how it performs? aka.. 64bit CPU vs 32bit only CPU? The larger L1 cach vs smaller L1 cach. A HTT vs 800 FSB? How can it really be fair?
Kinda like comparing Fords to Chevy's
Used to be off the line.. ford would win. In the long run, Chevy would win.
Ideally, benchmarks should be run stock and overclocked, also with different RAM combinations. Basically it is information the results of which enable you to select which components will work best for you, and at what likely speeds/performance will this rig run.
huh, I think that if average joe is buying a computer at best buy then he prb wont be comparing many benchmarks at all. Benchmarks are pretty much the domain of the gamer/hardcore junkie and most of us like that do not buy from a store, we build it ourselves.
This is naturally unless the gamer has more money than brains, or wants the ultimate bragging rights and buys from VoodooPC or something. Then they CAN get 'em oc'd as "stock".
I think ultimately comps are very much like cars. If I have a 69 charger with a 440 hemi all stock and want to race the dude with the 70 boss 429 mustang with a 429 cobrajet all stock that is fine, they are both beasts of their day and would be a sweet race. That does not negate a race between said mustang and a 69 vanilla camaro with a balanced/blueprinted/polished 350 ported to 383 with a blower/supercharger on a massive quad carb. That is still a valid race and can show what a tweaker can do with a cheaper car to beat (maybe) and/or compete with a more expensive "stock" car.
Same with comps. I think benchies are just that, kinda like races that show what each one is doing on the quarter mile so that you have something to measure to. This way you can go back to your garage and mess with the timing, fuel mix etc and get that last single digit horsepower bump and go back and take top spot on the leader boards.
I think the hardware sites should benchmark stock and oced to the max on an aftermarket cooler.
Stock is for users who WORK with the hardware and therefore are not necessarily interested in OCing.
OCed is for enthusiasts who are quite interested in that lovely OCing =)
See, I agree with the stock speed tests however, OC'ing usually brings different results.
Depending on what board you use, chipset, RAM, PSU, Cooler Etc. you can and will get very different results in most cases which is where I draw the line on this.
Yes, its fair to test two high end chips against each other at stock speeds because that's how they come from the manufacturer. OC'ing can become very unfair in benchmarks. Besides, in all of the benchmarks where they test the EE vs. the FX60, they had to OC the EE to like 4.26 or higher just for it to be competitive and beat the 60 by a narrow margin. I think that's absurd if you ask me. Why should you have to clock a high end Intel chip that costs more money that high to be competitive?
I say these things now because I've seen and used machines with the EE clocked at 5Ghz or higher. They still just don't seem as zippy as the AMDs do. Not to mention, i'm using AMD now and I find them to be great performers.
personally I would not go so far as to say that it is unfair. Even here at tom's the benchmarks show the EE and the FX at stock and oc. Those that actually read the charts would see how much each oc is and where the margin is. Those that don't just go to the conclusion and read that the EE had to be oc'd enough to bake a cake to compete and conclude that the FX is stil better.
I still think the average joe, even if he does check out benchmarks simply reads the conclusion and goes on. I like seeing the stock AND oc charts. Naturally for me stock holds more sway b/c oc is not a sure thing... what you get may not go that high. Most hardware geeks know this too i think. (could be wrong though, have been before...)
I do think it would be best to do both, however, I think in the absence of one, they should be done at stock.
Heres my reasoning:
Most people buy their chips at stock, and don't overclock. Granted a lot of them don't look at benchmarks either.
I think its ridiculous that you have to buy a product, then modify its running parameters in order to beat or even keep up with the competition.
A lot of people buy their PCs from shops or "computer guys" that DO look at the benchmarks, and while a lot of these shops are biased, the benchmarks are still helpful. These consumers decisions are ultimately based on benchmarks (though they may not know it), and most of these are not OCed either.
The problem with the overclock benchmarks is it's bound to be unfair to one chip mfg or the other - unless they get a few hundred chips from each (never going to happen) and pick the cream of the crop from each side (better would be something like an 90% cutoff - i.e. the point where 90% of the chips can run at that speed or better). It's a crap shoot as to how well any ONE chip will OC. The ONLY thing guaranteed to someone buying a CPU is that it will run at the stock rated speed.
That said - I really don't care if they OC and show the benches, as long as they show benches at stock speed too. But if the reviewer implies that the average Joe will get the same OC he got, without doing a very large sampling of representative chips - it's just plain misleading. How many posts have we all seen - "why can't I OC my <whatever> over xGHz"?