How do benchmark tests work?

[narwhals88]

I'm looking at building a new computer and i'm leaning more towards AMD for a cost friendly build
But i checked out cpubenchmark.net to look at scores
and the AMD phenom 2 x6 scored a good deal lower than the i5 2500k
How is that possible when they have similar clock speeds
and the AMD has 6 cores where as the intel has 4 cores with no hyper threading
is it merely because of the 3.7 Ghz speed boost on intel's part?
i want to make a medium to good gaming comp, and i'm wondering if i should spend a hundred or so extra bucks and go with intel
Heres the cpubenchmark site that i'm looking at
http://www.cpubenchmark.net/cpu.php?cpu=Intel+Core+i5-2500K+@+3.30GHz

 

[amenx]

2 different architectures. Clock speeds are not comparable. More cores is not enough to compensate for the differences either. Its why the AMD x6 is priced lower than the i5.

 

[narwhals88]

2 different architectures. Clock speeds are not comparable. More cores is not enough to compensate for the differences either. Its why the AMD x6 is priced lower than the i5.

and how exactly do the different architectures result in different benchmark testings?
and is the benchmark comparable to normal computer use? internet, email, games, video editing, etc?

 

[TuxDave]

and how exactly do the different architectures result in different benchmark testings?
and is the benchmark comparable to normal computer use? internet, email, games, video editing, etc?

Each benchmark tries to model overall system performance by performing different tasks. But if you do video editing, you should go look for a video editing benchmark to see how well it does. Same with gaming, etc....

As for the different architectures, imagine two ways of doing the same thing. I want to send a message to my friend. I can write a letter and send it to him or I can send him a text message. One is clearly faster than the other and the benchmark on "who can send the message the fastest" will reflect that.

 

[narwhals88]

Each benchmark tries to model overall system performance by performing different tasks. But if you do video editing, you should go look for a video editing benchmark to see how well it does. Same with gaming, etc....

As for the different architectures, imagine two ways of doing the same thing. I want to send a message to my friend. I can write a letter and send it to him or I can send him a text message. One is clearly faster than the other and the benchmark on "who can send the message the fastest" will reflect that.

but wouldn't both amd and intel want to have the fastest architecture? like, why would one pick postal mail over texting when texting is clearly superior?
And wouldn't that be reflected in the clock speed?

 

[TuxDave]

but wouldn't both amd and intel want to have the fastest architecture? like, why would one pick postal mail over texting when texting is clearly superior?
And wouldn't that be reflected in the clock speed?

What if one of them knew how to text and the other one didn't?

Edit: Missed your second question. As for a non-tech analogy for clock speed, imagine an Olympic athlete vs a computer nerd. If they were to do the same thing the same way, the faster one (higher clock speed) would always be faster.

But if the nerd is texting and the athlete is running down the street to deliver the message, you can get the task done faster without having a higher "clock speed" and so you'll end up seeing the guy, who looks slower, get things done faster.

 

[Blitzvogel]

The results can vary, depending on how well the program/benchmark utilizes specific architectures. In general, comparing core vs core at same clock on let's say an AMD Phenom II vs Intel i series core, things can vary by almost a full 100% in Intel's favor, though generally it's around 20-50% I would say, and depends on the program/game of course. It's part of why even a 6 core Phenom II hardly holds a candle to a quad-core i5 or i7 in many tests, and in others just gets dominated. Right now Intel has the best architecture overall on the market for the most part, and it's more power efficient to boot. AMD has been trying to play catch up since 2006, so generally AMD competes on price-per-performance.

As for general computer use like internet, email, basic media, the experience remains generally similar as long as you have a decent amount of capability in your machine in the first place, like a modern higher speed dual core, 4 GB for Windows 7, etc. Gaming, video editing, and benchmarking are where things get skewed, though gaming usually "maxes" out the graphics performance before the CPU when running graphic settings at full tilt. There are plenty of titles out there that do test the hell out of the CPU though and having a better one will help in those instances of course.

A cheaper i7 series CPU, 4 GB of DDR3 RAM, and an Nvidia GTX 560 I would say is a good setup at the moment without getting too expensive. You could go AMD and save some money, but performance overall won't be as good in CPU intensive workloads and video editing/encode.

 

[narwhals88]

What if one of them knew how to text and the other one didn't?

They should know though, its no secret right? i mean AMD can go buy an intel chip, and reverse engineer it into their own. and vice versa.
I'm not really understanding this....

The results can vary, depending on how well the program/benchmark utilizes specific architectures. In general, comparing core vs core at same clock on let's say an AMD Phenom II vs Intel i series core, things can vary by almost a full 100% in Intel's favor, though generally it's around 20-50% I would say, and depends on the program/game of course. It's part of why even a 6 core Phenom II hardly holds a candle to a quad-core i5 or i7 in many tests, and in others just gets dominated.

As for general computer use like internet, email, basic media, the experience remains generally similar as long as you have a decent amount of capability in your machine in the first place, like a modern higher speed dual core. Gaming, video editing, and benchmarking are where things get skewed, though gaming usually "maxes" out the graphics performance before the CPU when running graphic settings at full tilt. There are plenty of titles out there that do test the hell out of the CPU though and having a better one will help in those instances of course.

But why do they vary in intel's favor
clock speed is how quickly a processor can do a task correct? so if the clock speeds and core numbers were the same, how does one over take the other??

 

[TuxDave]

They should know though, its no secret right? i mean AMD can go buy an intel chip, and reverse engineer it into their own. and vice versa.
I'm not really understanding this....

It IS a secret and while you can try to make educated guesses on how they do thing "well, if they got this done in 100 cycles, the only way they could've done it is if maybe they do BLAH". But without espionage, there's no way to know exactly how the other company does it.

 

[narwhals88]

It IS a secret and while you can try to make educated guesses on how they do thing "well, if they got this done in 100 cycles, the only way they could've done it is if maybe they do BLAH". But without espionage, there's no way to know exactly how the other company does it.

but what about this. clock speed measures how quickly it can do the task
If one is at 3.3 Ghz and the other is at 3.3 Ghz
they should be equal correct?
it should be more like sending an email vs sending a text to someone that has both email and texting on their smart phone
they both should get the job done at the same time
so why does intel always get a higher score in spite of both processors having similar if not the same clock speed?

 

[TuxDave]

but what about this. clock speed measures how quickly it can do the task
If one is at 3.3 Ghz and the other is at 3.3 Ghz
they should be equal correct?
it should be more like sending an email vs sending a text to someone that has both email and texting on their smart phone
they both should get the job done at the same time
so why does intel always get a higher score in spite of both processors having similar if not the same clock speed?

No, clock speed does not measure how fast a CPU can get a task done. It's a physical number like "steps per second'. Two guys can take the same # of steps per second but if one takes bigger steps, he'll get to the destination in fewer steps and hence less time. Clocks don't mean anything else.

 

[Blitzvogel]

They should know though, its no secret right? i mean AMD can go buy an intel chip, and reverse engineer it into their own. and vice versa.
I'm not really understanding this....



But why do they vary in intel's favor
clock speed is how quickly a processor can do a task correct? so if the clock speeds and core numbers were the same, how does one over take the other??

According to this site:.

"In a computer, the clock cycle is the time between two adjacent pulses of the oscillator that sets the tempo of the computer processor. The number of these pulses per second is known as the clock speed, which is generally measured in Mhz (megahertz, or millions of pulses per second) and lately even in Ghz (gigahertz, or billions of pulses per second). The clock speed is determined by a quartz-crystal circuit, similar to those used in radio communications equipment. Some processors execute only one instruction per clock cycle. More advanced processors, described as superscalar, can perform more than one instruction per clock cycle. The latter type of processor gets more work done at a given clock speed than the former type. Similarly, a computer with a 32-bit bus will work faster at a given clock speed than a computer with a 16-bit bus. For these reasons, there is no simple, universal relation among clock speed, "bus speed," and millions of instructions per second (MIPS)."

It's what it can do per cycle that matters. Intel's architecture simply can do more per cycle than AMDs. While there are some advantages to the architecture in AMDs stuff, for the most part Intel's architecture has things engineered in a manner that makes it superior. There are numerous components in a CPU, and how those components are designed, set up and connected to the others impact the overall performance. In some benchmarks, a chip may be far superior, in another benchmark, the exact same in performance to the competition, and in another may just be completely weaker. That's why multiple types of benchmarks are used to test the CPU at specific types of workloads and programs.

 

[paperwastage]

but wouldn't both amd and intel want to have the fastest architecture? like, why would one pick postal mail over texting when texting is clearly superior?
And wouldn't that be reflected in the clock speed?

// in the world of GPUs, nVidia and ATI BOTH have been caught trying to cheat the benchmarks (like detecting the benchmark running and altering some parameters to increase the score)

http://www.google.com/search?client...rceid=opera&ie=utf-8&oe=utf-8&channel=suggest

anyways, i dont believe AMD or Intel does these underhanded maneuvers