Are all MHz created equal?

[nicklmg]

NCIX: http://bit.ly/1O9BbgE

Amazon: http://geni.us/3kFq

 

Does an improvement in IPC (instructions per clock) actually amount to an imporvment in overall performance?

 

[blu4]

Thanks Nick. I am surprised how the Q6600 stacked up against the new Intel beast. Still, it's outdated now and even if you are on a budget, you should consider at least an i3/i5.

[RevoltTrain]

You had a bit of a typo there Nick

[MegaDave91]

I would say no-- being hit in the nads MHz.

[Darkr]

I don't think do...

Thanks nick!

[Faceman]

I like how LMG has been tiptoeing around how poor the FX series is with their series of videos, without even coming close to mentioning it by name.  Well done.

[Opcode]

I like how LMG has been tiptoeing around how poor the FX series is with their series of videos, without even coming close to mentioning it by name.  Well done.

Could you be any more biased, well done.

[Faceman]

Could you be any more biased, well done.

There is nothing to be biased about.  It is proven.  They have been tapdancing around the subject without addressing it head on in their recent videos.  The how many cores does gaming need, and now this.  Maybe it isn't their intention, but I see it as such.

[zappian]

This video is silly .

The gaming tests are at 4k where the çpu doesn't matter much and the games are GPU bound as hell .

So no a 775 quad çore inst "almost" the same as a 5960x , at 4k sure , at 1080 p never.

[Opcode]

There is nothing to be biased about.  It is proven.  They have been tapdancing around the subject without addressing it head on in their recent videos.  The how many cores does gaming need, and now this.  Maybe it isn't their intention, but I see it as such.

You're mistaken, these video's aren't being made to thrash Bulldozer performance but to distinguish the difference between MHz and IPC along with their influence. The fact that you came in here guns blazing thrashing AMD with stupid remarks like "LMG made this video to hint at shit FX performance" is about as biased (or actually fanboy) as it gets. The last line of your edited post only validates what's already been clarified along with contradicting your previous post.

[sidas]

I actually would like someone to do an AMD equivalent test.

[Praxis727]

Good Vid Nick!!!  Very creative.  I would have used 2 processors that have been closer together rather then that big of a gap but I am sure there would have been similar results. 2700k -3770k 4770k  I would be interested to see the differences between IVY/Sandy Bridge and haswell as well as Devils Canyon even.  MHZ Battle of the architectures. 

[Dabombinable]

You're mistaken, these video's aren't being made to thrash Bulldozer performance but to distinguish the difference between MHz and IPC along with their influence. The fact that you came in here guns blazing thrashing AMD with stupid remarks like "LMG made this video to hint at shit FX performance" is about as biased (or actually fanboy) as it gets. The last line of your edited post only validates what's already been clarified along with contradicting your previous post.

Ok, why should he share is opinion about how bad the FX line of CPU's are? Considering Wolfdale based CPU's smack the crap out of an FX 8350 in tasks that require up to but no more than 4 threads (at approximately the same frequency). And Wolfdale is nearly 4 years older than Bulldozer.

[Opcode]

Ok, why should he share is opinion about how bad the FX line of CPU's are? Considering Wolfdale based CPU's smack the crap out of an FX 8350 in tasks that require up to but no more than 4 threads (at approximately the same frequency). And Wolfdale is nearly 4 years older than Bulldozer.

Opinions are one thing but throwing words in LMG's mouth is another thing entirely that I don't think the team would appreciate.

[Dabombinable]

Opinions are one thing but throwing words in LMG's mouth is another thing entirely that I think the team wouldn't appreciate.

True, they've already done their piece on AMD.

[Opcode]

Try, they've already done their piece on AMD.

Try what? The lack of proper context is amusing.

 

P.S. My 6800k will smash any Wolfdale you can find. Just simple facts dual cores don't cut it anymore and the G3258 is a fine testament to that.

[harrynowl]

I assume what @Faceman was trying to say in a not-so-subtle way is that some of the recent videos answer some of the common questions/points of disagreement within the CPU/Motherboards subforum, typically surrounding AMD CPUs and their gaming performance.

[Opcode]

I assume what @Faceman was trying to say in a not-so-subtle way is that some of the recent videos answer some of the common questions/points of disagreement within the CPU/Motherboards subforum, typically surrounding AMD CPUs and their gaming performance.

There's certainly better ways of putting it than causing unrest in the forum. AMD is up there next to Haswell core performance and people still reference Bulldozer.

[alpenwasser]

I think shall stay on my dual-socket LGA1366 setup for a while yet.

[spwath]

You're mistaken, these video's aren't being made to thrash Bulldozer performance but to distinguish the difference between MHz and IPC along with their influence. The fact that you came in here guns blazing thrashing AMD with stupid remarks like "LMG made this video to hint at shit FX performance" is about as biased (or actually fanboy) as it gets. The last line of your edited post only validates what's already been clarified along with contradicting your previous post.

He never said the message the video just to hint at bad performance.

They made the video, and then hinted at it

[Opcode]

He never said the message the video just to hint at bad performance.

They made the video, and then hinted at it

He stated the video was a reflection of LMG simply hinting at FX poor performance. I've watched the video a few times now and not once was the FX series mentioned especially in the unsettle way he implies. The video can be used to understand FX poor performance even though they have all them Migglehurts but certainly doesn't imply it.

[Dabombinable]

Try what? The lack of proper context is amusing.

 

P.S. My 6800k will smash any Wolfdale you can find. Just simple facts dual cores don't cut it anymore and the G3258 is a fine testament to that.

Typo-True

PS. Wolfdale based, including quad cores.

[husochan]

Ok this Video is BS. It was quiet obvious how they would perform.

More interesting would be 2500K vs 4690K and guess what in games int wont be existent. There was no real progress over the last 3 Intel generations that could be relevant for your audience. In the CPU department we will mostly become more efficient and that is about it. Good for Intel that their sales don't rely too much on PC enthusiasts wallets, because performance wise it became quiet boring. So GPUs is where its at and please don't throw your money away

[OliverMD]

For those interested, IPC is related to a computer architecture field of thought called RISC (Reduced instruction set computers). A RISC based processor has a small amount of very simple instructions, this allows it to pipeline (overlap instructions).

 

For some background, an instruction (think a line of assembler) has to go through various stages of processing. Fore example it needs to be fetched from memory or cache, then the cpu needs to figure out what it means, perform the action, write it back to memory and so on and so forth. These are all different execution phases of the instruction. In a modern processor the pipeline depth is around 15, although it peaked at 31 with the Pentium 4. The pipeline depth means how many phases each instruction has for it to be processed. These phases can be overlapped, so 1 instruction can be decoded while another is executed. It follows that there are places in the processor for dealing with each phase of an instruction. The 'goal' is to fill up all of these so that you have many instructions being processed at the same time. This is all on 1 core.

 

Of course not all instructions need to go through all of these phases, so they can exit early, meaning more that 1 instruction per cycle is completed. The real time sink is memory access as this happens over many clock cycles, depending on which memory needs accessing (L1, L2, L3, RAM, HDD).

 

However it should be noted that not all processors are RISC, nor are they the opposite CISC, they are a mixture of the 2.

[ObeseWalrus]

Liked for GMM reference.