Choosing a Xeon Proc. - Cores or Ghz?

Considering a CPU upgrade and would like to hear some opinions -
(best if backuped with facts of course).
 
The choice is between a
2.4ghz/4 core/10M cache
or
2.0ghz/6 core/15M cache
 
All pros and cons considered...
Which would be better? 

Really not germane to the question.
So let me give you more detail.

I have a dual LGA 2011 socket motherboard with one CPU.
I decided on the Xeon E5 processor over the i series because
of the difference in FSB in relationship to DMA.

I bought the entry level on this board because of price.
But companies tend to entice you with a base price and bump hard
as you increase components.
So I buy my 'add-ons' through my own sources.

So while I could max out the given specs of two 2.9ghz/8Core/20M cache that is far outside of my budget.
There are several levels in-between though.
And I'm looking at the lower end of those units available

I've presented the two options I'm considering for this.
I could just as simply and for even less money buy a matching CPU/fan
for the other socket (which is a 1.8/4C/10Mc).
It just seems that I should consider uping the ghz and/or cores if I'm already doing this and since it is possible within my range.

Nuclear? Well, it is the 21st century...

Xeons for any reason are an absolute waste of your money send the baord back and get a refund.

a single 3930K overclocked to 4.5GHz will kill that Xeon until you get into DUAL 3.1GHz.
no reason at all for dual Xeons with audio. maybe just maybe if you are using Vienna's Mir and 3/4 of thier library

benchmarks: 2nd graph down

http://www.adkproaudio.com/benchmarks.cfm

a fool and his money are soon parted.

Apple is dead.

as i SAID you need DUAL 3.1GHz to beat the 3930k and if you look at 32 buffer the more than double the cost dual Xeon system barely beats the 3930K

both the processors you mentioned would be dog slow and over priced and vastly outperformed by the 3930K

lets see 3930K $580
single 3.1GHz $2000ish
never mind the motherboard was also double the price
the 3930k tied the single @ 64 buffer 

to answer your question GHZ is almost always better than more cores in most things

OH MY dang son not too bright are you.

the 3.1GHz is the FASTEST 8 core dual xeon there is nothing faster made by intel right now. its the E5-2687W $2000 a peice.
so thats 16 cores @ 3.1GHz going against a 6 core system @ 4.5GHz

think that pretty much answers your original question
considering the single 8 core losses to the higher GHz 6 at low latency and 16 core can barely beat it

those little podunk processors you are looking 2.4GHz quad core (E5-2609) and 2GHz 6 core (E5-2620) are going to be dog slow
in fact slower than a socket 1155 3770 @ $290

what chart mine? maybe you cant read it right  there are no E5 2.7GHz Xeons on my chart
something else you are not taking into consideration i sell dual Xeons (mostly for video)
i make far more on a dual Xeon than a single system
yet here i am saying DONT WASTE YOUR MONEY.... 

2nd graph down

DUal 3.1GHz Xeon 16 cores is the fastest there is (blue)  these are E5
single 3.1 Xeon 8 core red  again E5  no  other Xeons
3930k @ 4.5GHz  green   i7 socket 2011

go down to 32 buffer..  enough said..

FYI ALL socket 2011 are the exact same silicon (processors)
the 3930k is the same as the 8 core with 2 cores turned off so they are the same.
stock 3930K runs @ 3.5 and turbos to 3.9GHz..

again a fool and his money.

2nd chart down ONLY
 not the 3rd one  the 2nd is done with a newer test vs the older test..
too hard to explain the issues
  beside look again @ 32buffer if you really wanted to use that chart..  3930k/60X is still winning over the 8 core 2.6GHz
 

Stipes, Scott knows what he is talking about as he builds professional DAWs for a living.

To answer your question: Which of those two chips would I buy, for my uses?  Answer: neither.  For sample streaming, the speed of both of them are too slow to be cost effective.  FWIW, the key developer/tester of MIR at Vienna Symphonic Libraries uses a 3930k, and that does quite well enough for him.

Beyond the obvious, the answer to your question is actually a question: What do you intend to do with the system? 

What you will need will vary depending on how many tracks you intend to record, if you are doing live recording.  In that case your interface matters a great deal.  If you are going do be doing heavy sample streaming, generally, a faster CPU will be needed.  Sample streaming = faster speeds; effects processing = more cores.

Anyway, if you want to buy one of those xeons, go ahead.  After all, it is your money.  It is just that for audio purposes, there are better options for much less money, except in rare situations.

Hey Stipes, don't take Scott's sunny demeanor too seriously, he really does mean well and was trying to give you helpful and informed advice even if it didn't come across that way (with the added bonus of typos from banging his noggin on his keyboard).

Anyway, the ground upon which your original query trods was already well tilled by Scott and Vin a while back elsewhere:

http://www.gearslutz.com/...5-xeon-benchmarks.html

http://www.gearslutz.com/...2600k-2700k-3930k.html

http://www.gearslutz.com/...i7-3930k-who-wins.html

Oh, I'm sure we can disagree on something Scott.

So, what's better, Skyline or Empress or Gold Star or...?

Ha, it's a chili fest!


Hehe, not my location, just some past history and attachments. I do remember the Great Cinci Chili War though... (been there, ate that)
Mmm, Montgomery Inn bbq too!


Where for chili in Ft. Myers? (I drop by there on occasion)


Knew of a good chili place in Old Town Alexandria VA, TX theme but they did some decent Cinci chili, Lone Star Cafe iirc. (I get around)
Prefer to make my own 5-way these days, lots of peppers and a touch of cinnamon (not big on chocolate in my chili). 

Too bad about the hot sauce. Great for the sinuses (and, so I'm told, when one has to eat one's hat). 

I'm sure he's a nice guy. However, he didn't start by trying to be helpful. He started by stating a position not relative to the question at hand and it took a few post to get him to climb off that soapbox and give an answer (well half a vague answer). If a 'fool and his money are soon parted' then the person wanting to part them from it adopts the position to see everyone as a fool (such is the nature of those dependant on selling their wares).  His implications unfortunately say more about his character to me than what he attempts to impeach me with.

So in the many tests I've read that isn't ADK's it appears the 2.0-6c is a better choice than the 2.4-4c. (well he did say 'almost' when he declared speed over cores). However, I am looking more at a single 8c now.  So in that way we can consider him helpful in a sense. The 2.6 looks pretty good since you can pick one up around 15-1600.
It appears that gamers hate these xeon chips because of the belief that you can't overclock them. However, I have run accross some that do and tell me the only problem in doing so is that the lifespan of it is cut drastically. I don't overclock. In the past when I did all it seemed to do for me was increase the likelihood of system crashes. Video people like these chips because they are more stable. I like stable too. I also found information on the 3.1Xeon that says the boost to 3.8 diminishes to 3.3 when run in pairs. Perhaps that wasn't taken into consideration with some tests looking only for speed.

i7's are great chips. I don't see any dual motherboards for them though and then there is the archetectural difference concerning FSB/DMA. This is much like the difference in PCI and PCIe to me and seems to be where it's all headed.

I don't consider myself an expert in any of this though. So I present these ideas for open discussion and then when possible actually try them. Education is not cheap either. But what I'm most skeptical of are those that insist they're experts and gurus (and their supporters).

So in the many tests I've read that isn't ADK's it appears the 2.0-6c is a better choice than the 2.4-4c. (well he did say 'almost' when he declared speed over cores). However, I am looking more at a single 8c now.  So in that way we can consider him helpful in a sense. The 2.6 looks pretty good since you can pick one up around 15-1600.

The trouble is the tests made outside of our tiny audio community have very little to do with real time audio and will be misleading.  Server platforms are tested using server centric or HPC applications that naturally lend themselve to multi threaded use, so with those tests, more cores will win.  And in audio this same situation occurs only when the load is distributed among many light load plug ins, such as doing a large mix with many tracks and real time processing.  Using only a few very heavy soft synth loads is a different story.


Scott is honestly one of the best resources here, and while I agree his approach may have been abrasive, his advice should not be written off.  For MIDI / soft synth use, you will be better off with higher clock speed at the expense of core count.  I think ultimately his point was that a high clock i7 will outperform your multi socket server platform for way less money for what you are intending to use it for.

That implies one would know my entire use for it.
I've only given what is relative to this posit.

Speed performance in and of itself as an isolate might be impressive, but it is not the whole process.

Still, no one has offered any information on the correlation of dual chips. And it seems unfair to use any test designed for a single processor only as the benchmark for a pair. Seems to me that a different test is in order and then see how the single stacks up to that. But if we continue to use the same philosophy on something outside of that philosophy we can only find what we chose to find before we started the test.
So what intrigues me the most with these chips are that you can run them pairs and that you have direct memory access eliminating the fsb (or more accurately, that they do that already). Like pcie, this seems to make the case toward direct connections that earlier chips such as the i series lack.

I have a couple of powerful keyboards that run off little embedded Atom CPU's and this tells me, the chip is fundamentally not the problem. They use the Linux kernal though which seems to make it efficient. So while we race for faster pushed to its limits, there is another way that's simpler and more efficient. Seriously, as musicians we're stuck using gamer mentality parts with modified overlayed CAD programs running a kernel designed for banking.

Given these new technologies, it seems like an optimum time to step back out of this box they've put us all in. And if this out of the box view and off the wall questioning can do something, I'm willing to try.

first you are in an audio forum  so its all about audio
but ok you want video editing benchmarks i got those as well..

dude, fact is is your spent a ton of money for a very slow system and you will continue to debate this until you fingers fall off to make yourself believe or feel better about what you have done.

there is not a single test you can come up with including server tests that would show either of the dual xeon options beating a single affordable 6 core processor period.
i have been doing this 15 yrs (audio/video and in the past gaming and servers) as a biz and longer before that as an employee.