Anandtech useas oudated bechmark softwares on pupose(?)

[losbellos]

I would like to know that anandtech does uses the old bechmark softwares on purpose and/or they dont know that what are the limitations of these softwares.

Do you happen to know for example that cinebench 10 is unable to perform for example ambient occlusion pass multithreaded,so this is why intel processors are in advantage since hyperthreading can help in it a lot. Of course this doesnt happen in cinebench 11,5....

Same aplies to the 3dmax9 tests.

I would be so happy that anandtech would keep up the proffesional work on the tests, that at least would look after how these softwares are working, and why this processor is a lot faster than the other which is a lot faster in the newer version of the same benhmark software, for example cinebench, because the internet public starts to feel that you guys using such softwares on purpose where hyperthreading in advantage.... instead of testing true mulithreading performance.

For example finalRender,Vray is also have demo versions, that you can use to benchmark for today performance test and not 3-5+ years ago..

Can be this done please?


Thank you very much!

 

[zebrax2]

I would like to know that anandtech does uses the old bechmark softwares on purpose and/or they dont know that what are the limitations of these softwares.

Do you happen to know for example that cinebench 10 is unable to perform for example ambient occlusion pass multithreaded,so this is why intel processors are in advantage since hyperthreading can help in it a lot. Of course this doesnt happen in cinebench 11,5....

Same aplies to the 3dmax9 tests.

I would be so happy that anandtech would keep up the proffesional work on the tests, that at least would look after how these softwares are working, and why this processor is a lot faster than the other which is a lot faster in the newer version of the same benhmark software, for example cinebench, because the internet public starts to feel that you guys using such softwares on purpose where hyperthreading in advantage.... instead of testing true mulithreading performance.

For example finalRender,Vray is also have demo versions, that you can use to benchmark for today performance test and not 3-5+ years ago..

Can be this done please?


Thank you very much!

The reason probably is that changing their benchmark suite would mean retesting a lot of products for comparison purposes which consumes a lot of time.

 

[Abwx]

Worst is his so called Monte Carlo test under Excel...

This bench has to be really very very "optimised",
to the point that a Core2quad 6600 at 2.4ghz
outperform a 3.7ghz Phenom II X4...

 

[Idontcare]

The reason probably is that changing their benchmark suite would mean retesting a lot of products for comparison purposes which consumes a lot of time.

This.

It's simply a matter of convenience for aiding in the performance comparisons to systems tested in the past.

Worst is his so called Monte Carlo test under Excel...

This bench has to be really very very "optimised",
to the point that a Core2quad 6600 at 2.4ghz
outperform a 3.7ghz Phenom II X4...

What's so surprising?

Are you going to be like this when Llano is launched and all the on-die GPU comparisons are made to sandy-bridge? "ZOMFG, Llano, at much lower clockspeed to SB's IGP, trounces the SB IGP! This so called "GPU benchmark" is so totally over-optimized! For shame AT, for shame!".

 

[Abwx]

Are you going to be like this when Llano is launched and all the on-die GPU comparisons are made to sandy-bridge? "ZOMFG, Llano, at much lower clockspeed to SB's IGP, trounces the SB IGP! This so called "GPU benchmark" is so totally over-optimized! For shame AT, for shame!".

Your comparison is totally irrelevant..

Comparing Llano s IGP to Intel s HD2000/3000 is like comparing
a mono core CPU to a quad..
Performances from the start are largely different..

As for C2Q6600 outmatching a PII X4 980, it s not like the first
has a 50% performance advantage from the start since it s
comparables processors in matter of perfs/mhz...

 

[bunnyfubbles]

Worst is his so called Monte Carlo test under Excel...

This bench has to be really very very "optimised",
to the point that a Core2quad 6600 at 2.4ghz
outperform a 3.7ghz Phenom II X4...

OMG, a test where AMD does exceedingly poorly at, THROW IT OUT!!!

 

[Abwx]

OMG, a test where AMD does exceedingly poorly at, THROW IT OUT!!!

That AMD perform so poorly in this test is not due
to non optimised code in respect of the architecture
but a blatant optimisation to make it perform as poorly
as possible in a bench that will be retained as important
in productivity and thus influence negatively those who
are in charge of choosing the hardware for bureautics
purposes..

 

[bunnyfubbles]

That AMD perform so poorly in this test is not due
to non optimised code in respect of the architecture
but a blatant optimisation to make it perform as poorly
as possible in a bench that will be retained as important
in productivity and thus influence negatively those who
are in charge of choosing the hardware for bureautics
purposes..

conspiracy!

 

[sm625]

I can understand the monte carlo performance. No doubt some high-up from Goldamn Sachs or Citigroup got together with Otellini on the golf course one day and said "can you add some special optimizations for monte carlo? It would mean millions more in (skimmed off the taxpayer and/or our own clients) profits." AMD doesnt have the resources to dedicate an entire team just for optimizing for a certain piece of software. Although they NEED to be doing that with DirectX.

ps Did you catch the Goldman Sachs typo? 😀

 

[Cogman]

That AMD perform so poorly in this test is not due
to non optimised code in respect of the architecture
but a blatant optimisation to make it perform as poorly
as possible in a bench that will be retained as important
in productivity and thus influence negatively those who
are in charge of choosing the hardware for bureautics
purposes..

Prove it. The fact that AMD sucks at this test is NOT proof that the test was "optimized" so that amd would suck.

 

[sm625]

How long will it be before microsoft uses the new intel compiler on a new version of excel? Only then can we know for sure, unless somebody knows exactly what it is doing and writes an openoffice plugin which can than be recompiled specifically for an AMD cpu. But of course if microsoft never switches to the new compiler then you might have another conspiracy...

 

[Cogman]

How long will it be before microsoft uses the new intel compiler on a new version of excel? Only then can we know for sure, unless somebody knows exactly what it is doing and writes an openoffice plugin which can than be recompiled specifically for an AMD cpu. But of course if microsoft never switches to the new compiler then you might have another conspiracy...

What compiler do you think Microsoft uses? You do realize they write their OWN compilers, right? Microsoft's compiler is well documented to be pretty fair regardless of platform.

Why would it be a conspiracy for microsoft to continue using their own compilers?

 

[Lonyo]

This.

It's simply a matter of convenience for aiding in the performance comparisons to systems tested in the past.



What's so surprising?

Are you going to be like this when Llano is launched and all the on-die GPU comparisons are made to sandy-bridge? "ZOMFG, Llano, at much lower clockspeed to SB's IGP, trounces the SB IGP! This so called "GPU benchmark" is so totally over-optimized! For shame AT, for shame!".

They update their GPU benchmark suite regularly and even did pretty much a total clearout recently.
Retesting some samples of older CPUs shouldn't be too difficult if they decided to do an update. A dual core Core 2, a quad Core 2 etc (only one of each).
You don't need to retest absolutely everything.

 

[Phynaz]

Do you happen to know for example that cinebench 10 is unable to perform for example ambient occlusion pass multithreaded,so this is why intel processors are in advantage since hyperthreading can help in it a lot.

It's not multi-threaded so a CPU with Symmetric Multi Threading runs it faster??

 

[Voo]

How long will it be before microsoft uses the new intel compiler on a new version of excel? Only then can we know for sure, unless somebody knows exactly what it is doing and writes an openoffice plugin which can than be recompiled specifically for an AMD cpu. But of course if microsoft never switches to the new compiler then you might have another conspiracy...

Yeah because as is well known MS - the developers of their own C++ compiler and IDE - are using ICC. Sure it's known which VC version is used for which OS release, but then nobody would stop the office team from deciding to use a completely different compiler for fun - because as we all know that's a trivial and completely problemfree process! So you wouldn't care to proof your FUD? Otherwise it just looks like you've got no idea what you're talking about..

 

[A5]

That AMD perform so poorly in this test is not due
to non optimised code in respect of the architecture
but a blatant optimisation to make it perform as poorly
as possible in a bench that will be retained as important
in productivity and thus influence negatively those who
are in charge of choosing the hardware for bureautics
purposes..

Quit manually inserting line breaks into your posts. The forum does it for you and makes it much easier to read.

Also, the people who make purchasing decisions at most companies don't read AT.

 

[996GT2]

That AMD perform so poorly in this test is not due
to non optimised code in respect of the architecture
but a blatant optimisation to make it perform as poorly
as possible in a bench that will be retained as important
in productivity and thus influence negatively those who
are in charge of choosing the hardware for bureautics
purposes..

So AMD's current generation (Phenom II, Athlon II) sucks. Get over it.

Even in all other benchmarks that aren't "purposefully optimized for Intel", latest Phenom II X4s can't even match Intel's 4 year old Core 2 Quad clock for clock. Yeah, some real innovation there 🙄

 

[losbellos]

It's not multi-threaded so a CPU with Symmetric Multi Threading runs it faster??

No, its all about hyper-threading. Hyper threading can work on that taks much better while amd-s are stucked there. Cinema 4d 11,5 doesnt not have that limitation that for the current core's AO passes it have to wait until the code runs through all the other core's passes. Doing this of course in small chunks.

Anyway, why use 3-5 year old bench softwares?

I pretty much care only about rendering performance. Since x6 showed up situation changed pretty much. Still above 2500k, an about the same level as the core i7-920@3.2 Ghz. which is not that bad... and AMD is 33% percent cheaper. Overclocked runs very conviencing.

So Amd is not that bad, take a look at here for example.
http://www.cbscores.com/

By the way cinebench runs better on intel procs! 😀

 

[piesquared]

Using the "it's just too darn hard to rebench all those processors" is nothing but a convenient excuse. Are benchmarks not supposed to represent current performance? Using 4 year old software is nothing but pandering to whoever pads their wallet the biggest. Who cares how 4 year old hardware performs on 4 year old software. We've seen those numbers, that's right, 4 years ago. If current software changes the order in those lovely graphs, it stands to reason it would be fairly important to consider. But we wouldn't want that would we!

 

[Cogman]

Using the "it's just too darn hard to rebench all those processors" is nothing but a convenient excuse. Are benchmarks not supposed to represent current performance? Using 4 year old software is nothing but pandering to whoever pads their wallet the biggest. Who cares how 4 year old hardware performs on 4 year old software. We've seen those numbers, that's right, 4 years ago. If current software changes the order in those lovely graphs, it stands to reason it would be fairly important to consider. But we wouldn't want that would we!

For CPUs, 4 year old software is still very relevant. CPU extensions are usually just not all that important (other than for fringe applications). There really isn't a whole lot that changes with CPUs instructions that makes a 4 year old CPU test much different from one compiled yesterday.

GPUs are a different story. They do actually see a lot of changes which makes old GPU testing software somewhat irrelevant.

 

[drizek]

It is kind of annoying. I remember when I was benchmarking the iPad I had to go and get older versions of the software so I could compare the scores with anands. I think it isn't unreasonable to run the test twice, once with the old, standard version, and one with the current version.

And it's not about optimizations changing over software versions, its about the numbers from each being incompatible and uncomparable.

 

[dac7nco]

I pretty much care only about rendering performance. Since x6 showed up situation changed pretty much. Still above 2500k, an about the same level as the core i7-920@3.2 Ghz. which is not that bad... and AMD is 33% percent cheaper. Overclocked runs very conviencing.

An overclocked 2500K will handily beat an overclocked 1100T in absolutely anything you wish to do. Don't hate on Intel for being bigger, smarter and better; I was poor once too, buddy... I know it sucks.

Daimon

 

[Arkaign]

I would like to know that anandtech does uses the old bechmark softwares on purpose and/or they dont know that what are the limitations of these softwares.

Do you happen to know for example that cinebench 10 is unable to perform for example ambient occlusion pass multithreaded,so this is why intel processors are in advantage since hyperthreading can help in it a lot. Of course this doesnt happen in cinebench 11,5....

Same aplies to the 3dmax9 tests.

I would be so happy that anandtech would keep up the proffesional work on the tests, that at least would look after how these softwares are working, and why this processor is a lot faster than the other which is a lot faster in the newer version of the same benhmark software, for example cinebench, because the internet public starts to feel that you guys using such softwares on purpose where hyperthreading in advantage.... instead of testing true mulithreading performance.

For example finalRender,Vray is also have demo versions, that you can use to benchmark for today performance test and not 3-5+ years ago..

Can be this done please?


Thank you very much!

AT Cpu benches are very in-depth. All that you would find with some newer obscure stuff like Vray/Finalrender is that when everything is running correctly, the PhII is within a few % of C2Q clock for clock. All of which are pretty much blown to bits by SB/etc.

Tune back in with BD launch.

 

[losbellos]

An overclocked 2500K will handily beat an overclocked 1100T in absolutely anything you wish to do. Don't hate on Intel for being bigger, smarter and better; I was poor once too, buddy... I know it sucks.

Daimon

Thats gross man, what about the athlonxp series?
Why dont you overclcok the 1100T too?

 

[Phynaz]

No, its all about hyper-threading. Hyper threading can work on that taks much better while amd-s are stucked there.

Hyperthreading is Symmetric Multi Threading. HT is Intels' brand name for SMT.

What did you think it is? And how did you think it helps in single threaded applications?