AMD FX WinRAR monster? - No way!

[thebanik]

Massman: It is simillary as this idea:

IMC of SB 1155 is 1333 MHz, no more (if u want believe me, download datasheet). FX has 1866 MHz default IMC. So than why people tetsing both platforms with 1600 MHz usually?

 

Isnt that a contradictory statement??? Reviewers change these settings to show that given equal opportunity what a platform/cpu/another hardware can do, if we were to run everything on manufacturer suggested settings, then we should go ahead and publish the results from Reviewers guides that manufacturers send us.

 

 

Or looks at some boards and Turbo ratio. SOme boards for SB-dt has classic Intel turbo at default setings-, but some boards performed more better, because here is modified turbo (some ASUS and Gigabyte boards). If is somethink default, I testing with this default, no changing it for "review". For interesting value I can show people some tweaks etc, of course, this is right. This thought is not about my AMD or Intel fans....

 

Again, if you are not letting your readers know about the difference in boards default settings and how it would affect its users, then you are not doing your job properly???? If you are comparing motherboards, and you let the default turbo mode ON where some are running 3.9Ghz and some at 3.8Ghz then its not really a fair review.

[flanker]

the banik: To boards reviews...Why not? 99.9% users dont overclocking. Only the buy some components and put it in. Yes, this is reality. If one board will have diferent turbo at default BIOS setings, why not? It is added value of product. This added value in diferent turbo can have a bit negative side in power consumption. Of course, but good reviewer in review must this all aspect explain. This is my assumption.

 

Massman:

http://www.intel.com/content/www/us/en/processors/core/2nd-gen-core-desktop-vol-1-datasheet.html part of 1.2.1 (of course, SB-E has not 1333, but 1600 MHz)

 

http://www.amd.com/us/products/desktop/processors/amdfx/Pages/amdfx-key-architectural-features.aspx

or details here: http://support.amd.com/us/Processor_TechDocs/49686.pdf

 

 

easy ,-)

Edited November 23, 2011 by flanker

[Le085]

Hi guys.Thanks to this thread we are gone a little bit more in depth to Core Parking: http://www.xtremehardware.it/eng-reviews/eng-reviews/core-parking-on-windows-seven-winrar-performance-with-sandy-bridge--201111226092/

 

We measured also power consumption under load and we have seen no changes at all with core parking enabled or disabled under load (WinRAR)

This will proabably mean that leaving Core parking enabled could lead to higher power consumption under winrar because the system take several more time to compress the file (while the consumption does not change at all).

Maybe this is because HT is an almost virtual feature and the real core is still enabled and working.

 

Edited November 23, 2011 by Le085

[flanker]

very interesting with power consumption.

[Massman]

I agree with Flanker when he says that the reviewer needs to clearly state how results are obtained or if there's a reason why a certain product is faster or slower. For instance, for the Sandy Bridge-E mainboard comparisons, the reviewer should always point out that one board is faster than the other because it has a fixed turbo frequency of 39x. That fixed turbo frequency is a feature of the board's BIOS and if this feature is enabled on all the retail boards, it's important for end-users to know.

 

The same goes for this registry tweak. An editor could write that Bulldozer appears to be quicker in WinRAR, but that it's due to a bug in windows causing the application not to make full use of all the hardware's features. And, that with a simple registry tweak, you can improve the performance easily.

 

Massman:

http://www.intel.com/content/www/us/en/processors/core/2nd-gen-core-desktop-vol-1-datasheet.html part of 1.2.1 (of course, SB-E has not 1333, but 1600 MHz)

 

easy ,-)

 

Not so easy. I've been looking for information on the IMC/ring frequency for a long time now, but it's documented nowhere. I don't see anything in that particular section of the documentation that specifies the IMC frequency running at DDR3-1333. All I see are references to transfer rates and DDR3 frequencies, with 1333MHz being the highest officially supported DRAM frequency on SB and DDR3-1600 being the highest officially supported on SB-E.

 

We don't know how fast the IMC is running. It can be 1:1 with the CPU frequency, it can be 1:1 with memory (although I seriously doubt that). It might be 2:3, maybe even 1:2 ... we can only guess.

[xoqolatl]

On Sandy Bridge μ-arch ring bus and last level cache run at core frequency, particularly at the same frequency as highest clocked core (confirmed by Intel on IDF). Note that this does not say anything about how the data crosses teh ringbus/IMC clock domains, or how fast is the IMC running. It could very well have a queue for data coming on and off the ring and run at a fraction of that speed.

[SoF]

nice find!