[Tom's] Normalized single-core CPU performance at 3GHz

[Ancalagon44]

Does no one else find it odd that K8 at 3GHz outperforms K10 at 3GHz according to this chart? Maybe it just happens in some isolated cases, but it is still odd.

I also noticed that, pretty concerning to be honest.

 

[jvroig]

Ok, then let me rephrase that.

I'm confused as to how the athlon IIs perform almost the same as the phenom IIs in this chart.

Wouldn't that mean they'd have to perform almost equally in multi threaded apps?

No, because that's what the L3 cache contributes the most for (multi-threaded scenarios). It may not make much sense for single-threaded apps (so you see only 4&#37, but as you've no doubt seen in reviews, the additional L3 in Phenom II's can give as much as 15 or 20% better performance in different scenarios compared to the L3-less Athlon II line.

 

[StrangerGuy]

This particular chart gives a nice summary: http://www.tomshardware.com/charts/...parison/All-time-based-values-added,2779.html. Overall, SB is the biggest jump (10%) in an Intel design since Pentium 4 to Conroe (an amazing 50%). Also, you'll see that AMD's newest CPUs still trail Conroe in single-core performance.

Try at least 100%. Every review shows a gimped 2M L2 1.83GHz E6300 besting a 3.73GHz P-EE.

 

[LOL_Wut_Axel]

Try at least 100%. Every review shows a gimped 2M L2 1.83GHz E6300 besting a 3.73GHz P-EE.

Remember that it has one additional core. For a comparison to be fair this way it'd have to be against a Pentium D, or with the Core 2 Duo with one disabled core against the Pentium 4.

 

[Termie]

Try at least 100%. Every review shows a gimped 2M L2 1.83GHz E6300 besting a 3.73GHz P-EE.

Remember that it has one additional core. For a comparison to be fair this way it'd have to be against a Pentium D, or with the Core 2 Duo with one disabled core against the Pentium 4.

Exactly - this is not a CPU vs. CPU comparison, it's a core vs. core comparison.

But...I actually wasn't sure how to represent that number when I posted it. I first wrote 100% and then I changed it to 50%. I think I'm having trouble with my basic math right now...in a time-based test like iTunes, how do you compute how much faster a lower score is? Here's the chart: http://www.tomshardware.com/charts/x86-core-performance-comparison/iTunes-9.0.3.15,2768.html. If the P4 is at 227 seconds and the 2600k is at 80 seconds, is the 2600k 64% faster, or is it 129% faster, or is it something else?

Put another way, if one chip cuts the processing time exactly in half, is it 50% faster or 100% faster? That's just about what Conroe did to P4.

 

[LOL_Wut_Axel]

Exactly - this is not a CPU vs. CPU comparison, it's a core vs. core comparison.

But...I actually wasn't sure how to represent that number when I posted it. I first wrote 100% and then I changed it to 50%. I think I'm having trouble with my basic math right now...in a time-based test like iTunes, how do you compute how much faster a lower score is? Here's the chart: http://www.tomshardware.com/charts/x86-core-performance-comparison/iTunes-9.0.3.15,2768.html. If the P4 is at 227 seconds and the 2600k is at 80 seconds, is the 2600k 64% faster, or is it 129% faster, or is it something else?

Put another way, if one chip cuts the processing time exactly in half, is it 50% faster or 100% faster? That's just about what Conroe did to P4.

129% faster. You always take the faster CPU as the baseline.

If it cuts processing time in half it's twice as fast (eg. X=50 seconds and Y=100 seconds; X is 2x faster, or 100% faster).

 

[zsdersw]

Netburst was a complete failure.

Not really.. Northwood-C was quite good.

 

[Topweasel]

Not really.. Northwood-C was quite good.

The problem isn't its competitiveness at certain points. Its that Intel spent a butt load of money developing that to last them till right around now. They got lucky that the Pentium M, which was developed because was worried about Transmeta, was so stellar and clocked as high as it did. That allowed them the ability to wait out Conroe, which still was originally part of the Pentium M development line. It wasn't till Nahelam that it Intel had the opportunity steer the ship back to a more Desktop oriented design and re-establish a proper road map.

 

[zsdersw]

The problem isn't its competitiveness at certain points. Its that Intel spent a butt load of money developing that to last them till right around now. They got lucky that the Pentium M, which was developed because was worried about Transmeta, was so stellar and clocked as high as it did. That allowed them the ability to wait out Conroe, which still was originally part of the Pentium M development line. It wasn't till Nahelam that it Intel had the opportunity steer the ship back to a more Desktop oriented design and re-establish a proper road map.

Banias was Intel being typical Intel; "Only the paranoid survive". It was their "Plan B".. not so much luck as intentional choice.

 

[soccerballtux]

Instead, couldn't you keep the 3ghz core clock and overclock the CPU-NB to ~2700mhz? I feel like this alone would match wolfdale. I can't really find many useful benchmarks on this.

yeah in games it's about 10% faster when you OC the NB to 2.5ghz.
there are benches somewhere with this. Hard to find them.
Explanation is that the faster response from the L3 cache helps significantly with a L2 cache miss. In other words it's if it's not in L2 it's very likely that it's in L3.

 

[hamunaptra]

This is a pointless test really. I mean seriously....the cpu world is going more multithreaded year after year, software and OS's are being more multithreaded optimized - its the way of the future. In light of this, to run a battery of tests against CPU's that arent necessarily designed for single threaded performance is just dumb.

AMD has broken away from trying to achieve max throughput on a single core for a LONG time, hence why they were the first into dual core, quad core, hex core and now octo core for consumer market (on a single die).

I mean sure, having less cores but are more powerful obviously shows the efficiency of intels cores in current day applications. But, to keep tryin for the fastest throughput single core is kinda , IMO, not so futuristic thinking.
I think the future will be more cores upon more cores, whoever embraces this first will reap the benefits.
Tweaking a core around max x86 throughput...only gets you so far due to the inherent restrictions in the ISA itself. More cores will ultimately be the answer in the future of x86.

Anyways, it just bugs me sites continually go on to find fault in AMD's uarch and how "weak" it is....

If they wanna run some real tests...fucking run multiple stress tests programs simultaneously while running games and other things....on a modern intel / AMD platform...see who has a better experience at handling all these tasks at the same time.
In that case, Ive seen AMD come much further ahead then intel.
But its odd no sites test like that anymore. They run one damn benchmark at a time... sometimes it aint about how fast a machine can perform a single task, sometimes its about how much total work can a machine achieve given multiple tasks at the same time.

 

[lol123]

Single-threaded performance is still king both with regard to gaming and in many enterprise applications, so comparing it across different processor generations isn't exactly pointless.

 

[LOL_Wut_Axel]

This is a pointless test really. I mean seriously....the cpu world is going more multithreaded year after year, software and OS's are being more multithreaded optimized - its the way of the future. In light of this, to run a battery of tests against CPU's that arent necessarily designed for single threaded performance is just dumb.

AMD has broken away from trying to achieve max throughput on a single core for a LONG time, hence why they were the first into dual core, quad core, hex core and now octo core for consumer market (on a single die).

I mean sure, having less cores but are more powerful obviously shows the efficiency of intels cores in current day applications. But, to keep tryin for the fastest throughput single core is kinda , IMO, not so futuristic thinking.
I think the future will be more cores upon more cores, whoever embraces this first will reap the benefits.
Tweaking a core around max x86 throughput...only gets you so far due to the inherent restrictions in the ISA itself. More cores will ultimately be the answer in the future of x86.

Anyways, it just bugs me sites continually go on to find fault in AMD's uarch and how "weak" it is....

If they wanna run some real tests...fucking run multiple stress tests programs simultaneously while running games and other things....on a modern intel / AMD platform...see who has a better experience at handling all these tasks at the same time.
In that case, Ive seen AMD come much further ahead then intel.
But its odd no sites test like that anymore. They run one damn benchmark at a time... sometimes it aint about how fast a machine can perform a single task, sometimes its about how much total work can a machine achieve given multiple tasks at the same time.

I guess you could make the argument that Intel has been stuck with Quad-Cores for under $500 since 2006. We're definitely moving towards more multi-threading, though. Most important programs now support it except for audio encoding.

This is just to compare architectural differences at the most basic level without taking into account new features the newer CPUs may have.

 

[LOL_Wut_Axel]

Single-threaded performance is still king both with regard to gaming and in many enterprise applications, so comparing it across different processor generations isn't exactly pointless.

LMAO! All games from 2009 to now make use of at least three threads and enterprise applications are almost always the first to take advantage of multi-threading since they'll get workloads done quicker, and more efficiently.

Have you been living under a rock or what? Single-threaded applications are now pretty much limited to only audio encoding.

 

[VirtualLarry]

This is a pointless test really. I mean seriously....the cpu world is going more multithreaded year after year, software and OS's are being more multithreaded optimized - its the way of the future. In light of this, to run a battery of tests against CPU's that arent necessarily designed for single threaded performance is just dumb.

I agree that they should also included multi-threaded tests, that show how well a CPU handles multithreading (IPC - Interprocess Communication), message-passing, cache-line-sharing, etc.

I mean sure, having less cores but are more powerful obviously shows the efficiency of intels cores in current day applications. But, to keep tryin for the fastest throughput single core is kinda , IMO, not so futuristic thinking.
I think the future will be more cores upon more cores, whoever embraces this first will reap the benefits.
Tweaking a core around max x86 throughput...only gets you so far due to the inherent restrictions in the ISA itself. More cores will ultimately be the answer in the future of x86.

But there's a limitation on how well multi-cores can scale, due to Ahmdal's law, so single-threaded performance is ALWAYS important.

Anyways, it just bugs me sites continually go on to find fault in AMD's uarch and how "weak" it is....

That's because it IS WEAK. Why do you think AMD charges less for their CPUs than Intel? Just because AMD is such a nice company? No! They have to, they have less performance.

 

[LOL_Wut_Axel]

I agree that they should also included multi-threaded tests, that show how well a CPU handles multithreading (IPC - Interprocess Communication), message-passing, cache-line-sharing, etc.

But there's a limitation on how well multi-cores can scale, due to Ahmdal's law, so single-threaded performance is ALWAYS important.


That's because it IS WEAK. Why do you think AMD charges less for their CPUs than Intel? Just because AMD is such a nice company? No! They have to, they have less performance.

IPC means Instructions Per Cycle.

/Just sayin'

 

[hamunaptra]

I agree that they should also included multi-threaded tests, that show how well a CPU handles multithreading (IPC - Interprocess Communication), message-passing, cache-line-sharing, etc.

But there's a limitation on how well multi-cores can scale, due to Ahmdal's law, so single-threaded performance is ALWAYS important.


That's because it IS WEAK. Why do you think AMD charges less for their CPUs than Intel? Just because AMD is such a nice company? No! They have to, they have less performance.

LOL AMD uarch is NOT WEAK. Theres a reason that there exists an unfortunate unquantifiable difference between intel and AMD's CPU's perceived smoothness in operation when doing multiple tasks at once. Ive have seen many accounts and heard many accounts of people stating their AMD system just feels and plays smoother than a comparable intel one. Hence why its unfortunate so many sites dont do true multi tasking reviews...and its nearly impossible to test the "smoothness" of a system.
Just because a system can pull 100fps+ in a game doesnt mean its running "smoother" than a 40fps system. I and many others can attest to this.

Also, much code is unfortunately written for optimization on a intel uarch rather than being equally as fair to a AMD CPU. Hell, half the benchmarks in existence are rigged for this.
AMD gets bad press, gets bad bench results all because of this. THAT is why they have to sell their stuff for less.

Now, I am not saying AMD's uarch is the one and only, all powerful, alpha and omega of CPU's...they arent as strong as intels in some areas, specially some technical areas on paper. But, some areas they are equal or better, especially in code optimized for AMD - which is RARE.
Imagine a world where code was written specifically tailored to their uarch, much like most code in existence is for intel uarch. The playing field would look SIGNIFICANTLY different. If code was completely unbiased between one or the other, it would still look different then it does now.

Given the way intel slants everyone in the industry towards optimizing for their uarch, I think AMD is doing a damn good job at keepin in the race.

 

[Phynaz]

LMAO! All games from 2009 to now make use of at least three threads

Three CPU heavy threads?

 

[VirtualLarry]

IPC means Instructions Per Cycle.

/Just sayin'

Sure, wise guy. But in the software field, IPC does mean InterProcess Communication.

Search MSDN sometime.

http://msdn.microsoft.com/en-us/library/system.runtime.remoting.channels.ipc.aspx

 

[VirtualLarry]

Theres a reason that there exists an unfortunate unquantifiable difference between intel and AMD's CPU's perceived smoothness in operation when doing multiple tasks at once. Ive have seen many accounts and heard many accounts of people stating their AMD system just feels and plays smoother than a comparable intel one.

If it's unquantifiable, then it falls under the heading of "faith". So you AMD true believers get together, and keep the faith alive.

 

[zsdersw]

LOL AMD uarch is NOT WEAK. Theres a reason that there exists an unfortunate unquantifiable difference between intel and AMD's CPU's perceived smoothness in operation when doing multiple tasks at once. Ive have seen many accounts and heard many accounts of people stating their AMD system just feels and plays smoother than a comparable intel one. Hence why its unfortunate so many sites dont do true multi tasking reviews...and its nearly impossible to test the "smoothness" of a system.
Just because a system can pull 100fps+ in a game doesnt mean its running "smoother" than a 40fps system. I and many others can attest to this.

Also, much code is unfortunately written for optimization on a intel uarch rather than being equally as fair to a AMD CPU. Hell, half the benchmarks in existence are rigged for this.
AMD gets bad press, gets bad bench results all because of this. THAT is why they have to sell their stuff for less.

Now, I am not saying AMD's uarch is the one and only, all powerful, alpha and omega of CPU's...they arent as strong as intels in some areas, specially some technical areas on paper. But, some areas they are equal or better, especially in code optimized for AMD - which is RARE.
Imagine a world where code was written specifically tailored to their uarch, much like most code in existence is for intel uarch. The playing field would look SIGNIFICANTLY different. If code was completely unbiased between one or the other, it would still look different then it does now.

Given the way intel slants everyone in the industry towards optimizing for their uarch, I think AMD is doing a damn good job at keepin in the race.

Oy, here we go with the whole compiler/optimization tempest-in-a-teapot.

It's a nugget of fiction that crops up from time to time.

 

[magomago]

Can someone explain to me why this measurement is relevant? And I'm not talking just about 'single cpu performance', I'm talking about the methodology used (adding up times from a bunch of tests?) and why this is a valid way to test CPU performance.

On top of it, is this test a true measure of single core performance when there are other factors like L2 and L3 cache? Should those be crippled down to the lowest common denominator? Would that still be an accurate measure of single core performance?

The point I'm trying to make is that these are retarded conclusions based on a metric that has no real explanation or validity beyond "add up timesss!". No shhiiiiittttt Intel is faster --> real benchmarks consisting of actual applications in actual usage scenarios will show that.

But I don't understand the logic of how these measurements translate to actual performance at all, and I don't think anyone could even begin to explain it foundationally: using these charts, X6 1100T is about 72% the speed of the i2600k . We'd also say that a i2600K is 3x faster than a pentium 4. Does any of this make sense? NO. To make it even more ridiculous, they threw video encoding in on it...which these days want more and more cores. Finally, multi core performance is different from single core performance and programs are shifting to multi core environments....so exactly what are we really getting out of it??? That programs which are limited to single cores will run faster? How about we look at the actual programs as opposed to a confusing mish mash of zipping up a bunch of files, running 3D mark, rendering an image, and running synethic benchmarks to come up with an even more confusing number who metrics cannot be properly understood

 

[tijag]

I'm confused, how do the i5-2500k and the i7-2600k perform differently if these benchmarks are supposed to show 1 thread, at the exact same frequency?

 

[lol123]

LMAO! All games from 2009 to now make use of at least three threads and enterprise applications are almost always the first to take advantage of multi-threading since they'll get workloads done quicker, and more efficiently.

Have you been living under a rock or what? Single-threaded applications are now pretty much limited to only audio encoding.

Games have been made to utilize more than one core (for example to make physics calculations), but the main game thread is still a single-threaded application. There is little that can be done to change that fact because of the nature of what a game is, where everything that happens and which has to be computed and rendered is dependent on what the player decides to do next. Therefore it's not a problem that can be effectively parallellized (whereas the rendering of each individual frame can).

When it comes to enterprise applications you should give some thought to why IBM's POWER CPUs sell like hotcakes despite their high price, enormous power draw and low number of cores compared to for example Sun's (and now Oracle's) T3 processors - because of their superior single-threaded performance. Nothing else.

 

[footballrunner800]

I'm confused, how do the i5-2500k and the i7-2600k perform differently if these benchmarks are supposed to show 1 thread, at the exact same frequency?

Remember the 2600k has 2mb more of L3 cache.