Overclock.net banner

1 - 13 of 13 Posts

·
Banned
Joined
·
288 Posts
Discussion Starter #1
I don't get it, why do review sites still insist on using low-quality settings to benchmark CPUs, and make the ludicrous statement that it reflects the true power of a processor?

Look at this:

From the recent LegionHardware review of the Core i7.



The Core i7 totally tools on all the rest of the processors right? especially the measly Phenom, outscoring it by over 60fps at the same clockspeeds... Right?

Wrong.


Next we see here, at High quality settings, which is what 99.9% of the people on this website would be running, The core i7 920 actually loses to the very same Phenom that it just smashed in the low quality tests.

Why do people still think its alright to judge a processor when they use settings that no one uses... especially people who spend 300 dollars on Quad-Core Nehalems.

It's ridiculous.

High:
http://www.legionhardware.com/document.php?id=786&p=9
Low:
http://www.legionhardware.com/document.php?id=786&p=10

Edit: there seems to be a temporary problem with the legion hardware servers, so the images might not show for you, this should be resolved in time, but for the time being, just click on the links
 

·
Registered
Joined
·
3,150 Posts
Well,that's actually a VERY good point there.
If you gonna review for enthusiasts do it on the high settings where it hurts the GPU.
 

·
Registered
Joined
·
2,530 Posts
But the whole point is that at very low settings, the GPU does very little work, and the CPU has to do more work.
 

·
Premium Member
Joined
·
5,275 Posts
I agree. Max or nothing. Compare them all with the same video card.

lol
 

·
Registered
Joined
·
2,530 Posts
It is just another way to bench a CPU.

It is no more relevant than SuperPi.
 

·
Banned
Joined
·
288 Posts
Discussion Starter #6
Quote:


Originally Posted by TEntel
View Post

But the whole point is that at very low settings, the GPU does very little work, and the CPU has to do more work.

I know what the point is, but what does it matter if a CPU can score 2905724958 fps points at low settings when the processors main goal is to run high settings well, and it can't?

If its an enthusiast processor, give me enthusiast settings.

Quote:


Originally Posted by TEntel
View Post

It is just another way to bench a CPU.

It is no more relevant than SuperPi.

Why would game benchmarks not be relevant?
 

·
Registered
Joined
·
2,530 Posts
It doesn't matter anymore than how many seconds it takes to calculate pi to the millionth decimal point.

To some people, they aren't relevant. Not everyone plays games.
My point is that the test is simply another way to compare CPUs.

It doesn't matter what the tests may actually mean, all the matters is how it performs relative to others.
 

·
Banned
Joined
·
288 Posts
Discussion Starter #8
Quote:


Originally Posted by TEntel
View Post

It doesn't matter anymore than how many seconds it takes to calculate pi to the millionth decimal point.

To some people, they aren't relevant. Not everyone plays games.
My point is that the test is simply another way to compare CPUs.

It doesn't matter what the tests may actually mean, all the matters is how it performs relative to others.

exactly, but you're still missing my point.

In an enthusiast setting (remember, a core i7 costs upwards of $300) why are low settings even considered? its not an accurate gauge of a processor's power whatsoever, because the people who buy $300 processors and DO play games are going to be led astray by the results that many of these review sites are publishing.

A CEO who is buying these processors for his company isn't going to even look at Far Cry 2 tests, he's going to look at SiSandra and SuperPi, and he will get some hinkling of the processors performance based on those numbers, because all he cares about is its number crunching capabilities.

So what I'm trying to say is, they shouldn't use low settings benchmarks in reviews, maybe as a supplement sure, but reviews suggesting that people buy a Core i7, for example, based on the notion that its 100% faster in games than a Phenom when the only tests they have done are low settings tests, are somewhat outrageous to me, because they will only end up making their loyal readers buy the wrong product.
 

·
Registered
Joined
·
2,530 Posts
These review sites need to have something for everybody.

There are a lot of "standard" benchmark tests that are run, and they mean different things to different people.

But regardless of all that, the fact is that the higher the resolution, the less the CPU matters.

At the lowest resolution possible, the CPU is the most important variable.

Regardless of the settings that may be used in real life, many gamers will find those benchmarks very interesting. They often point to other side effects and quirks of the CPU.

And heck, I think I make a good example.
I consider myself a computer enthusiast.
I'm a Computer Engineering Major.
I don't play games.
I would buy a $300 CPU without hesitation.

Even though those low resolution settings don't pertain 100% to gaming, they do hint at performance in other areas.

If one were to be considering dropping all that cash on a CPU, it would be smart to read a lot of reviews, and check out a bunch of different benchies...

I dunno.
It is very late at night.
I'm rambling and I forget what my point was.
 

·
Banned
Joined
·
288 Posts
Discussion Starter #10
Quote:


Originally Posted by TEntel
View Post

These review sites need to have something for everybody.

There are a lot of "standard" benchmark tests that are run, and they mean different things to different people.

But regardless of all that, the fact is that the higher the resolution, the less the CPU matters.

At the lowest resolution possible, the CPU is the most important variable.

Regardless of the settings that may be used in real life, many gamers will find those benchmarks very interesting. They often point to other side effects and quirks of the CPU.

And heck, I think I make a good example.
I consider myself a computer enthusiast.
I'm a Computer Engineering Major.
I don't play games.
I would buy a $300 CPU without hesitation.

Even though those low resolution settings don't pertain 100% to gaming, they do hint at performance in other areas.

If one were to be considering dropping all that cash on a CPU, it would be smart to read a lot of reviews, and check out a bunch of different benchies...

I dunno.
It is very late at night.
I'm rambling and I forget what my point was.


That's all well and good, like I said:

Quote:


they shouldn't use low settings benchmarks in reviews, maybe as a supplement sure...

My only gripe is with review sites that ONLY show the low settings results because they think that those are the only ones that matter.

By all means, they should keep the low settings scores for the curious people like you
, but I think it should be a requirement to show the high settings too.
 

·
I might have tacos tonite
Joined
·
22,339 Posts
I don't think you seem to understand why they use low settings?

The CPU does the exact same work at the low settings. When you test a CPU, you want to be JUST testing the cpu, not the other components of the system.

If they ran the settings at all super-high max, every single CPU benchmark would turn partly into a graphics card battle, and would not give accurate results. However, any settings affecting CPU should be left on, like physics settings, ect.

The buyer should be aware that their system will only be as fast as the slowest component in any given application.
 

·
Banned
Joined
·
288 Posts
Discussion Starter #12
Quote:

Originally Posted by Crazy9000 View Post
I don't think you seem to understand why they use low settings?

The CPU does the exact same work at the low settings. When you test a CPU, you want to be JUST testing the cpu, not the other components of the system.

If they ran the settings at all super-high max, every single CPU benchmark would turn partly into a graphics card battle, and would not give accurate results. However, any settings affecting CPU should be left on, like physics settings, ect.

The buyer should be aware that their system will only be as fast as the slowest component in any given application.
if all of those setups use the same graphics card then, and they are all somehow limited by it, why aren't their scores all exactly the same?

If low settings were a show of true power, than the Core i7 would beat the Phenom at high settings, but it doesn't, because they are not.

I know exactly what the theory behind low settings is, I'm just saying its flawed.
 

·
Registered
Joined
·
2,530 Posts
Quote:


Originally Posted by Henry_John_Scott
View Post

If low settings were a show of true power, than the Core i7 would beat the Phenom at high settings, but it doesn't, because they are not.


The problem with your argument is that you are basing it on an assumption: the benchmarks on this site are inherently wrong.

The reason why you don't like them is because you don't like what you see.
If everything was consistent across the board, you would have no complaints.

The more likely scenario is that the benches are exposing a flaw in the i7, not that the well established benches are themselves flawed.
 
1 - 13 of 13 Posts
Top